# HG changeset patch # User mcharles # Date 1409911930 14400 # Node ID e8e6b962c1f237f48005243aa2ca06b6e2a3f3d2 # Parent edddaa8ab855f8ee6c66bdb3befba12c5301f52f Uploaded diff -r edddaa8ab855 -r e8e6b962c1f2 rapsodyn/Galaxy-Workflow-Rapsodyn_v1.34.ga --- a/rapsodyn/Galaxy-Workflow-Rapsodyn_v1.34.ga Thu Aug 21 08:39:15 2014 -0400 +++ /dev/null Thu Jan 01 00:00:00 1970 +0000 @@ -1,692 +0,0 @@ -{ - "a_galaxy_workflow": "true", - "annotation": "", - "format-version": "0.1", - "name": "Rapsodyn v1.34", - "steps": { - "0": { - "annotation": "", - "id": 0, - "input_connections": {}, - "inputs": [ - { - "description": "", - "name": "READ1" - } - ], - "name": "Input dataset", - "outputs": [], - "position": { - "left": 200, - "top": 200 - }, - "tool_errors": null, - "tool_id": null, - "tool_state": "{\"name\": \"READ1\"}", - "tool_version": null, - "type": "data_input", - "user_outputs": [] - }, - "1": { - "annotation": "", - "id": 1, - "input_connections": {}, - "inputs": [ - { - "description": "", - "name": "READ2" - } - ], - "name": "Input dataset", - "outputs": [], - "position": { - "left": 200, - "top": 285 - }, - "tool_errors": null, - "tool_id": null, - "tool_state": "{\"name\": \"READ2\"}", - "tool_version": null, - "type": "data_input", - "user_outputs": [] - }, - "2": { - "annotation": "", - "id": 2, - "input_connections": {}, - "inputs": [ - { - "description": "", - "name": "ASSEMBLY" - } - ], - "name": "Input dataset", - "outputs": [], - "position": { - "left": 200, - "top": 370 - }, - "tool_errors": null, - "tool_id": null, - "tool_state": "{\"name\": \"ASSEMBLY\"}", - "tool_version": null, - "type": "data_input", - "user_outputs": [] - }, - "3": { - "annotation": "", - "id": 3, - "input_connections": {}, - "inputs": [ - { - "description": "", - "name": "DUBIOUS POSITION" - } - ], - "name": "Input dataset", - "outputs": [], - "position": { - "left": 200, - "top": 455 - }, - "tool_errors": null, - "tool_id": null, - "tool_state": "{\"name\": \"DUBIOUS POSITION\"}", - "tool_version": null, - "type": "data_input", - "user_outputs": [] - }, - "4": { - "annotation": "", - "id": 4, - "input_connections": { - "input_file": { - "id": 0, - "output_name": "output" - } - }, - "inputs": [], - "name": "FASTQ Groomer", - "outputs": [ - { - "name": "output_file", - "type": "fastqsanger" - } - ], - "position": { - "left": 428, - "top": 200 - }, - "post_job_actions": {}, - "tool_errors": null, - "tool_id": "toolshed.g2.bx.psu.edu/repos/devteam/fastq_groomer/fastq_groomer/1.0.4", - "tool_state": "{\"__page__\": 0, \"input_file\": \"null\", \"input_type\": \"\\\"illumina\\\"\", \"__rerun_remap_job_id__\": null, \"options_type\": \"{\\\"options_type_selector\\\": \\\"basic\\\", \\\"__current_case__\\\": 0}\", \"chromInfo\": \"\\\"/home/galaxy/galaxy-dist/tool-data/shared/ucsc/chrom/?.len\\\"\"}", - "tool_version": "1.0.4", - "type": "tool", - "user_outputs": [] - }, - "5": { - "annotation": "", - "id": 5, - "input_connections": { - "input_file": { - "id": 1, - "output_name": "output" - } - }, - "inputs": [], - "name": "FASTQ Groomer", - "outputs": [ - { - "name": "output_file", - "type": "fastqsanger" - } - ], - "position": { - "left": 428, - "top": 331 - }, - "post_job_actions": {}, - "tool_errors": null, - "tool_id": "toolshed.g2.bx.psu.edu/repos/devteam/fastq_groomer/fastq_groomer/1.0.4", - "tool_state": "{\"__page__\": 0, \"input_file\": \"null\", \"input_type\": \"\\\"illumina\\\"\", \"__rerun_remap_job_id__\": null, \"options_type\": \"{\\\"options_type_selector\\\": \\\"basic\\\", \\\"__current_case__\\\": 0}\", \"chromInfo\": \"\\\"/home/galaxy/galaxy-dist/tool-data/shared/ucsc/chrom/?.len\\\"\"}", - "tool_version": "1.0.4", - "type": "tool", - "user_outputs": [] - }, - "6": { - "annotation": "", - "id": 6, - "input_connections": { - "input_fasta": { - "id": 2, - "output_name": "output" - } - }, - "inputs": [], - "name": "fastaGroomerForMakeBlastdb", - "outputs": [ - { - "name": "output_fasta", - "type": "fasta" - } - ], - "position": { - "left": 428, - "top": 462 - }, - "post_job_actions": {}, - "tool_errors": null, - "tool_id": "fastaGroomerForMakeBlastdb", - "tool_state": "{\"input_fasta\": \"null\", \"__rerun_remap_job_id__\": null, \"chromInfo\": \"\\\"/home/galaxy/galaxy-dist/tool-data/shared/ucsc/chrom/?.len\\\"\", \"__page__\": 0}", - "tool_version": "0.01", - "type": "tool", - "user_outputs": [] - }, - "7": { - "annotation": "", - "id": 7, - "input_connections": { - "readtype|input_paired1": { - "id": 4, - "output_name": "output_file" - }, - "readtype|input_paired2": { - "id": 5, - "output_name": "output_file" - } - }, - "inputs": [], - "name": "Sickle", - "outputs": [ - { - "name": "output_single", - "type": "input" - }, - { - "name": "output_combo", - "type": "input" - }, - { - "name": "output_combo_single", - "type": "input" - }, - { - "name": "output_paired1", - "type": "fastq" - }, - { - "name": "output_paired2", - "type": "fastq" - }, - { - "name": "output_paired_single", - "type": "fastq" - } - ], - "position": { - "left": 756, - "top": 200 - }, - "post_job_actions": {}, - "tool_errors": null, - "tool_id": "toolshed.g2.bx.psu.edu/repos/nikhil-joshi/sickle/sickle/1.33", - "tool_state": "{\"__page__\": 0, \"length_threshold\": \"\\\"30\\\"\", \"no_five_prime\": \"\\\"False\\\"\", \"__rerun_remap_job_id__\": null, \"readtype\": \"{\\\"single_or_paired\\\": \\\"pe_sep\\\", \\\"input_paired1\\\": null, \\\"input_paired2\\\": null, \\\"__current_case__\\\": 2}\", \"trunc_n\": \"\\\"False\\\"\", \"qual_threshold\": \"\\\"30\\\"\"}", - "tool_version": "1.33", - "type": "tool", - "user_outputs": [] - }, - "8": { - "annotation": "", - "id": 8, - "input_connections": { - "input_file": { - "id": 6, - "output_name": "output_fasta" - } - }, - "inputs": [], - "name": "NCBI BLAST+ makeblastdb", - "outputs": [ - { - "name": "outfile", - "type": "data" - } - ], - "position": { - "left": 756, - "top": 494 - }, - "post_job_actions": {}, - "tool_errors": null, - "tool_id": "toolshed.g2.bx.psu.edu/repos/devteam/ncbi_blast_plus/ncbi_makeblastdb/0.1.00", - "tool_state": "{\"__page__\": 0, \"mask_data_file\": \"null\", \"input_file\": \"null\", \"dbtype\": \"\\\"nucl\\\"\", \"__rerun_remap_job_id__\": null, \"hash_index\": \"\\\"True\\\"\", \"tax\": \"{\\\"taxselect\\\": \\\"\\\", \\\"__current_case__\\\": 0}\", \"title\": \"\\\"\\\"\", \"chromInfo\": \"\\\"/home/galaxy/galaxy-dist/tool-data/shared/ucsc/chrom/?.len\\\"\", \"parse_seqids\": \"\\\"False\\\"\"}", - "tool_version": "0.1.00", - "type": "tool", - "user_outputs": [] - }, - "9": { - "annotation": "", - "id": 9, - "input_connections": { - "readtype|input_paired1": { - "id": 7, - "output_name": "output_paired1" - }, - "readtype|input_paired2": { - "id": 7, - "output_name": "output_paired2" - } - }, - "inputs": [], - "name": "Sickle", - "outputs": [ - { - "name": "output_single", - "type": "input" - }, - { - "name": "output_combo", - "type": "input" - }, - { - "name": "output_combo_single", - "type": "input" - }, - { - "name": "output_paired1", - "type": "fastq" - }, - { - "name": "output_paired2", - "type": "fastq" - }, - { - "name": "output_paired_single", - "type": "fastq" - } - ], - "position": { - "left": 1084, - "top": 200 - }, - "post_job_actions": {}, - "tool_errors": null, - "tool_id": "toolshed.g2.bx.psu.edu/repos/nikhil-joshi/sickle/sickle/1.33", - "tool_state": "{\"__page__\": 0, \"length_threshold\": \"\\\"30\\\"\", \"no_five_prime\": \"\\\"False\\\"\", \"__rerun_remap_job_id__\": null, \"readtype\": \"{\\\"single_or_paired\\\": \\\"pe_sep\\\", \\\"input_paired1\\\": null, \\\"input_paired2\\\": null, \\\"__current_case__\\\": 2}\", \"trunc_n\": \"\\\"True\\\"\", \"qual_threshold\": \"\\\"30\\\"\"}", - "tool_version": "1.33", - "type": "tool", - "user_outputs": [] - }, - "10": { - "annotation": "", - "id": 10, - "input_connections": { - "genomeSource|ownFile": { - "id": 2, - "output_name": "output" - }, - "paired|input1": { - "id": 9, - "output_name": "output_paired1" - }, - "paired|input2": { - "id": 9, - "output_name": "output_paired2" - } - }, - "inputs": [], - "name": "Map with BWA for Illumina", - "outputs": [ - { - "name": "output", - "type": "sam" - } - ], - "position": { - "left": 1412, - "top": 200 - }, - "post_job_actions": {}, - "tool_errors": null, - "tool_id": "toolshed.g2.bx.psu.edu/repos/devteam/bwa_wrappers/bwa_wrapper/1.2.3", - "tool_state": "{\"genomeSource\": \"{\\\"refGenomeSource\\\": \\\"history\\\", \\\"ownFile\\\": null, \\\"__current_case__\\\": 1}\", \"__page__\": 0, \"__rerun_remap_job_id__\": null, \"paired\": \"{\\\"input2\\\": null, \\\"sPaired\\\": \\\"paired\\\", \\\"input1\\\": null, \\\"__current_case__\\\": 1}\", \"params\": \"{\\\"__current_case__\\\": 0, \\\"source_select\\\": \\\"pre_set\\\"}\", \"suppressHeader\": \"\\\"False\\\"\", \"chromInfo\": \"\\\"/home/galaxy/galaxy-dist/tool-data/shared/ucsc/chrom/?.len\\\"\"}", - "tool_version": "1.2.3", - "type": "tool", - "user_outputs": [] - }, - "11": { - "annotation": "", - "id": 11, - "input_connections": { - "input_sam_file": { - "id": 10, - "output_name": "output" - } - }, - "inputs": [], - "name": "filtersam_mapped_and_unique", - "outputs": [ - { - "name": "output_file", - "type": "sam" - } - ], - "position": { - "left": 1676, - "top": 200 - }, - "post_job_actions": {}, - "tool_errors": null, - "tool_id": "filtersam_mapped_and_unique", - "tool_state": "{\"__page__\": 0, \"__rerun_remap_job_id__\": null, \"input_sam_file\": \"null\"}", - "tool_version": "0.01", - "type": "tool", - "user_outputs": [] - }, - "12": { - "annotation": "", - "id": 12, - "input_connections": { - "source|input1": { - "id": 11, - "output_name": "output_file" - }, - "source|ref_file": { - "id": 2, - "output_name": "output" - } - }, - "inputs": [], - "name": "SAM-to-BAM", - "outputs": [ - { - "name": "output1", - "type": "bam" - } - ], - "position": { - "left": 2004, - "top": 200 - }, - "post_job_actions": {}, - "tool_errors": null, - "tool_id": "toolshed.g2.bx.psu.edu/repos/devteam/sam_to_bam/sam_to_bam/1.1.4", - "tool_state": "{\"source\": \"{\\\"index_source\\\": \\\"history\\\", \\\"ref_file\\\": null, \\\"input1\\\": null, \\\"__current_case__\\\": 1}\", \"__rerun_remap_job_id__\": null, \"chromInfo\": \"\\\"/home/galaxy/galaxy-dist/tool-data/shared/ucsc/chrom/?.len\\\"\", \"__page__\": 0}", - "tool_version": "1.1.4", - "type": "tool", - "user_outputs": [] - }, - "13": { - "annotation": "", - "id": 13, - "input_connections": { - "reference_source|input_bams_0|input_bam": { - "id": 12, - "output_name": "output1" - }, - "reference_source|ref_file": { - "id": 2, - "output_name": "output" - } - }, - "inputs": [], - "name": "MPileup", - "outputs": [ - { - "name": "output_mpileup", - "type": "pileup" - }, - { - "name": "output_log", - "type": "txt" - } - ], - "position": { - "left": 2232, - "top": 200 - }, - "post_job_actions": {}, - "tool_errors": null, - "tool_id": "toolshed.g2.bx.psu.edu/repos/devteam/samtools_mpileup/samtools_mpileup/0.0.3", - "tool_state": "{\"__page__\": 0, \"genotype_likelihood_computation_type\": \"{\\\"genotype_likelihood_computation_type_selector\\\": \\\"do_not_perform_genotype_likelihood_computation\\\", \\\"__current_case__\\\": 1}\", \"__rerun_remap_job_id__\": null, \"advanced_options\": \"{\\\"max_reads_per_bam\\\": \\\"250\\\", \\\"advanced_options_selector\\\": \\\"advanced\\\", \\\"extended_BAQ_computation\\\": \\\"False\\\", \\\"region_string\\\": \\\"\\\", \\\"output_per_sample_strand_bias_p_value\\\": \\\"False\\\", \\\"minimum_base_quality\\\": \\\"30\\\", \\\"disable_probabilistic_realignment\\\": \\\"False\\\", \\\"skip_anomalous_read_pairs\\\": \\\"False\\\", \\\"minimum_mapping_quality\\\": \\\"0\\\", \\\"output_per_sample_read_depth\\\": \\\"False\\\", \\\"__current_case__\\\": 0, \\\"position_list\\\": null, \\\"coefficient_for_downgrading\\\": \\\"0\\\"}\", \"reference_source\": \"{\\\"ref_file\\\": null, \\\"reference_source_selector\\\": \\\"history\\\", \\\"input_bams\\\": [{\\\"__index__\\\": 0, \\\"input_bam\\\": null}], \\\"__current_case__\\\": 1}\", \"chromInfo\": \"\\\"/home/galaxy/galaxy-dist/tool-data/shared/ucsc/chrom/?.len\\\"\"}", - "tool_version": "0.0.3", - "type": "tool", - "user_outputs": [] - }, - "14": { - "annotation": "", - "id": 14, - "input_connections": { - "input_file": { - "id": 13, - "output_name": "output_mpileup" - } - }, - "inputs": [], - "name": "mpileupfilter", - "outputs": [ - { - "name": "output_file", - "type": "pileup" - } - ], - "position": { - "left": 2558, - "top": 200 - }, - "post_job_actions": {}, - "tool_errors": null, - "tool_id": "mpileupfilter", - "tool_state": "{\"__page__\": 0, \"input_file\": \"null\", \"chromInfo\": \"\\\"/home/galaxy/galaxy-dist/tool-data/shared/ucsc/chrom/?.len\\\"\", \"min_frequency\": \"\\\"0.8\\\"\", \"min_depth\": \"\\\"2\\\"\", \"__rerun_remap_job_id__\": null, \"min_forward_and_reverse\": \"\\\"0\\\"\", \"max_depth\": \"\\\"50\\\"\", \"min_distance\": \"\\\"0\\\"\"}", - "tool_version": "0.05", - "type": "tool", - "user_outputs": [] - }, - "15": { - "annotation": "", - "id": 15, - "input_connections": { - "input_listA": { - "id": 14, - "output_name": "output_file" - }, - "input_listB": { - "id": 3, - "output_name": "output" - } - }, - "inputs": [], - "name": "listfiltering", - "outputs": [ - { - "name": "output_file", - "type": "pileup" - } - ], - "position": { - "left": 2886, - "top": 200 - }, - "post_job_actions": {}, - "tool_errors": null, - "tool_id": "listfiltering", - "tool_state": "{\"__page__\": 0, \"__rerun_remap_job_id__\": null, \"nb_col\": \"\\\"2\\\"\", \"input_listB\": \"null\", \"input_listA\": \"null\", \"chromInfo\": \"\\\"/home/galaxy/galaxy-dist/tool-data/shared/ucsc/chrom/?.len\\\"\"}", - "tool_version": "0.01", - "type": "tool", - "user_outputs": [] - }, - "16": { - "annotation": "", - "id": 16, - "input_connections": { - "input_assembly_file": { - "id": 2, - "output_name": "output" - }, - "input_variant_file": { - "id": 15, - "output_name": "output_file" - } - }, - "inputs": [], - "name": "extractseq", - "outputs": [ - { - "name": "output_file", - "type": "fasta" - } - ], - "position": { - "left": 3214, - "top": 200 - }, - "post_job_actions": {}, - "tool_errors": null, - "tool_id": "extractseq", - "tool_state": "{\"__page__\": 0, \"input_variant_file\": \"null\", \"__rerun_remap_job_id__\": null, \"input_assembly_file\": \"null\", \"window_length\": \"\\\"50\\\"\", \"chromInfo\": \"\\\"/home/galaxy/galaxy-dist/tool-data/shared/ucsc/chrom/?.len\\\"\"}", - "tool_version": "0.03", - "type": "tool", - "user_outputs": [] - }, - "17": { - "annotation": "", - "id": 17, - "input_connections": { - "db_opts|histdb": { - "id": 8, - "output_name": "outfile" - }, - "query": { - "id": 16, - "output_name": "output_file" - } - }, - "inputs": [], - "name": "NCBI BLAST+ blastn", - "outputs": [ - { - "name": "output1", - "type": "tabular" - } - ], - "position": { - "left": 3542, - "top": 200 - }, - "post_job_actions": {}, - "tool_errors": null, - "tool_id": "toolshed.g2.bx.psu.edu/repos/devteam/ncbi_blast_plus/ncbi_blastn_wrapper/0.1.00", - "tool_state": "{\"evalue_cutoff\": \"\\\"0.001\\\"\", \"__page__\": 0, \"adv_opts\": \"{\\\"identity_cutoff\\\": \\\"0.0\\\", \\\"adv_opts_selector\\\": \\\"advanced\\\", \\\"ungapped\\\": \\\"False\\\", \\\"filter_query\\\": \\\"False\\\", \\\"word_size\\\": \\\"0\\\", \\\"__current_case__\\\": 1, \\\"parse_deflines\\\": \\\"False\\\", \\\"strand\\\": \\\"-strand both\\\", \\\"max_hits\\\": \\\"0\\\"}\", \"__rerun_remap_job_id__\": null, \"db_opts\": \"{\\\"db_opts_selector\\\": \\\"histdb\\\", \\\"subject\\\": \\\"\\\", \\\"histdb\\\": null, \\\"__current_case__\\\": 1, \\\"database\\\": \\\"\\\"}\", \"query\": \"null\", \"blast_type\": \"\\\"megablast\\\"\", \"output\": \"{\\\"out_format\\\": \\\"5\\\", \\\"__current_case__\\\": 3}\", \"chromInfo\": \"\\\"/home/galaxy/galaxy-dist/tool-data/shared/ucsc/chrom/?.len\\\"\"}", - "tool_version": "0.1.00", - "type": "tool", - "user_outputs": [] - }, - "18": { - "annotation": "", - "id": 18, - "input_connections": { - "input_blastxml_file": { - "id": 17, - "output_name": "output1" - }, - "input_variant_file": { - "id": 15, - "output_name": "output_file" - } - }, - "inputs": [], - "name": "mpileupfilteronblastxml", - "outputs": [ - { - "name": "output_file", - "type": "pileup" - } - ], - "position": { - "left": 3844, - "top": 200 - }, - "post_job_actions": {}, - "tool_errors": null, - "tool_id": "mpileupfilteronblastxml", - "tool_state": "{\"__page__\": 0, \"input_variant_file\": \"null\", \"__rerun_remap_job_id__\": null, \"input_blastxml_file\": \"null\", \"window_length\": \"\\\"50\\\"\", \"nb_mismatch_max\": \"\\\"3\\\"\", \"chromInfo\": \"\\\"/home/galaxy/galaxy-dist/tool-data/shared/ucsc/chrom/?.len\\\"\"}", - "tool_version": "0.03", - "type": "tool", - "user_outputs": [] - }, - "19": { - "annotation": "", - "id": 19, - "input_connections": { - "input_blast_filtered_file": { - "id": 18, - "output_name": "output_file" - }, - "input_list_filtered_file": { - "id": 15, - "output_name": "output_file" - }, - "input_mpileup_filtered_file": { - "id": 14, - "output_name": "output_file" - }, - "input_mpileup_row_file": { - "id": 13, - "output_name": "output_mpileup" - }, - "input_read1_row_file": { - "id": 0, - "output_name": "output" - }, - "input_read1_trimmed_file": { - "id": 7, - "output_name": "output_paired1" - }, - "input_read1_trimmednoN_file": { - "id": 9, - "output_name": "output_paired1" - }, - "input_read2_row_file": { - "id": 1, - "output_name": "output" - }, - "input_read2_trimmed_file": { - "id": 7, - "output_name": "output_paired2" - }, - "input_read2_trimmednoN_file": { - "id": 9, - "output_name": "output_paired2" - }, - "input_sam_filtered_file": { - "id": 11, - "output_name": "output_file" - }, - "input_sam_row_file": { - "id": 10, - "output_name": "output" - } - }, - "inputs": [], - "name": "rapsodyn_stats", - "outputs": [ - { - "name": "output_file", - "type": "txt" - } - ], - "position": { - "left": 4172, - "top": 200 - }, - "post_job_actions": {}, - "tool_errors": null, - "tool_id": "rapsodyn_stats", - "tool_state": "{\"input_read1_trimmed_file\": \"null\", \"input_read2_trimmednoN_file\": \"null\", \"input_read2_row_file\": \"null\", \"input_read2_trimmed_file\": \"null\", \"__page__\": 0, \"input_sam_row_file\": \"null\", \"__rerun_remap_job_id__\": null, \"input_mpileup_filtered_file\": \"null\", \"input_read1_trimmednoN_file\": \"null\", \"input_sam_filtered_file\": \"null\", \"input_read1_row_file\": \"null\", \"input_mpileup_row_file\": \"null\", \"input_list_filtered_file\": \"null\", \"input_blast_filtered_file\": \"null\"}", - "tool_version": "0.01", - "type": "tool", - "user_outputs": [] - } - } -} \ No newline at end of file diff -r edddaa8ab855 -r e8e6b962c1f2 rapsodyn/Galaxy-Workflow-Rapsodyn_v1.34_-_Filtering.ga --- a/rapsodyn/Galaxy-Workflow-Rapsodyn_v1.34_-_Filtering.ga Thu Aug 21 08:39:15 2014 -0400 +++ /dev/null Thu Jan 01 00:00:00 1970 +0000 @@ -1,374 +0,0 @@ -{ - "a_galaxy_workflow": "true", - "annotation": "", - "format-version": "0.1", - "name": "Rapsodyn v1.34 - Filtering", - "steps": { - "0": { - "annotation": "", - "id": 0, - "input_connections": {}, - "inputs": [ - { - "description": "", - "name": "BAM" - } - ], - "name": "Input dataset", - "outputs": [], - "position": { - "left": 200, - "top": 677.5 - }, - "tool_errors": null, - "tool_id": null, - "tool_state": "{\"name\": \"BAM\"}", - "tool_version": null, - "type": "data_input", - "user_outputs": [] - }, - "1": { - "annotation": "", - "id": 1, - "input_connections": {}, - "inputs": [ - { - "description": "", - "name": "ASSEMBLY" - } - ], - "name": "Input dataset", - "outputs": [], - "position": { - "left": 200, - "top": 762.5 - }, - "tool_errors": null, - "tool_id": null, - "tool_state": "{\"name\": \"ASSEMBLY\"}", - "tool_version": null, - "type": "data_input", - "user_outputs": [] - }, - "2": { - "annotation": "", - "id": 2, - "input_connections": {}, - "inputs": [ - { - "description": "", - "name": "DUBIOUS POSITION" - } - ], - "name": "Input dataset", - "outputs": [], - "position": { - "left": 200, - "top": 847.5 - }, - "tool_errors": null, - "tool_id": null, - "tool_state": "{\"name\": \"DUBIOUS POSITION\"}", - "tool_version": null, - "type": "data_input", - "user_outputs": [] - }, - "3": { - "annotation": "", - "id": 3, - "input_connections": { - "reference_source|input_bams_0|input_bam": { - "id": 0, - "output_name": "output" - }, - "reference_source|ref_file": { - "id": 1, - "output_name": "output" - } - }, - "inputs": [], - "name": "MPileup", - "outputs": [ - { - "name": "output_mpileup", - "type": "pileup" - }, - { - "name": "output_log", - "type": "txt" - } - ], - "position": { - "left": 428, - "top": 677.5 - }, - "post_job_actions": {}, - "tool_errors": null, - "tool_id": "toolshed.g2.bx.psu.edu/repos/devteam/samtools_mpileup/samtools_mpileup/0.0.3", - "tool_state": "{\"__page__\": 0, \"genotype_likelihood_computation_type\": \"{\\\"genotype_likelihood_computation_type_selector\\\": \\\"do_not_perform_genotype_likelihood_computation\\\", \\\"__current_case__\\\": 1}\", \"__rerun_remap_job_id__\": null, \"advanced_options\": \"{\\\"max_reads_per_bam\\\": \\\"250\\\", \\\"advanced_options_selector\\\": \\\"advanced\\\", \\\"extended_BAQ_computation\\\": \\\"False\\\", \\\"region_string\\\": \\\"\\\", \\\"output_per_sample_strand_bias_p_value\\\": \\\"False\\\", \\\"minimum_base_quality\\\": \\\"30\\\", \\\"disable_probabilistic_realignment\\\": \\\"False\\\", \\\"skip_anomalous_read_pairs\\\": \\\"False\\\", \\\"minimum_mapping_quality\\\": \\\"0\\\", \\\"output_per_sample_read_depth\\\": \\\"False\\\", \\\"__current_case__\\\": 0, \\\"position_list\\\": null, \\\"coefficient_for_downgrading\\\": \\\"0\\\"}\", \"reference_source\": \"{\\\"ref_file\\\": null, \\\"reference_source_selector\\\": \\\"history\\\", \\\"input_bams\\\": [{\\\"__index__\\\": 0, \\\"input_bam\\\": null}], \\\"__current_case__\\\": 1}\", \"chromInfo\": \"\\\"/home/galaxy/galaxy-dist/tool-data/shared/ucsc/chrom/?.len\\\"\"}", - "tool_version": "0.0.3", - "type": "tool", - "user_outputs": [] - }, - "4": { - "annotation": "", - "id": 4, - "input_connections": { - "input_fasta": { - "id": 1, - "output_name": "output" - } - }, - "inputs": [], - "name": "fastaGroomerForMakeBlastdb", - "outputs": [ - { - "name": "output_fasta", - "type": "fasta" - } - ], - "position": { - "left": 428, - "top": 889.5 - }, - "post_job_actions": {}, - "tool_errors": null, - "tool_id": "fastaGroomerForMakeBlastdb", - "tool_state": "{\"input_fasta\": \"null\", \"__rerun_remap_job_id__\": null, \"chromInfo\": \"\\\"/home/galaxy/galaxy-dist/tool-data/shared/ucsc/chrom/?.len\\\"\", \"__page__\": 0}", - "tool_version": "0.01", - "type": "tool", - "user_outputs": [] - }, - "5": { - "annotation": "", - "id": 5, - "input_connections": { - "input_file": { - "id": 3, - "output_name": "output_mpileup" - } - }, - "inputs": [], - "name": "mpileupfilter", - "outputs": [ - { - "name": "output_file", - "type": "pileup" - } - ], - "position": { - "left": 756, - "top": 677.5 - }, - "post_job_actions": {}, - "tool_errors": null, - "tool_id": "mpileupfilter", - "tool_state": "{\"__page__\": 0, \"input_file\": \"null\", \"chromInfo\": \"\\\"/home/galaxy/galaxy-dist/tool-data/shared/ucsc/chrom/?.len\\\"\", \"min_frequency\": \"\\\"0.8\\\"\", \"min_depth\": \"\\\"2\\\"\", \"__rerun_remap_job_id__\": null, \"min_forward_and_reverse\": \"\\\"0\\\"\", \"max_depth\": \"\\\"50\\\"\", \"min_distance\": \"\\\"0\\\"\"}", - "tool_version": "0.05", - "type": "tool", - "user_outputs": [] - }, - "6": { - "annotation": "", - "id": 6, - "input_connections": { - "input_file": { - "id": 4, - "output_name": "output_fasta" - } - }, - "inputs": [], - "name": "NCBI BLAST+ makeblastdb", - "outputs": [ - { - "name": "outfile", - "type": "data" - } - ], - "position": { - "left": 756, - "top": 808.5 - }, - "post_job_actions": {}, - "tool_errors": null, - "tool_id": "toolshed.g2.bx.psu.edu/repos/devteam/ncbi_blast_plus/ncbi_makeblastdb/0.1.00", - "tool_state": "{\"__page__\": 0, \"mask_data_file\": \"null\", \"input_file\": \"null\", \"dbtype\": \"\\\"nucl\\\"\", \"__rerun_remap_job_id__\": null, \"hash_index\": \"\\\"True\\\"\", \"tax\": \"{\\\"taxselect\\\": \\\"\\\", \\\"__current_case__\\\": 0}\", \"title\": \"\\\"\\\"\", \"chromInfo\": \"\\\"/home/galaxy/galaxy-dist/tool-data/shared/ucsc/chrom/?.len\\\"\", \"parse_seqids\": \"\\\"False\\\"\"}", - "tool_version": "0.1.00", - "type": "tool", - "user_outputs": [] - }, - "7": { - "annotation": "", - "id": 7, - "input_connections": { - "input_listA": { - "id": 5, - "output_name": "output_file" - }, - "input_listB": { - "id": 2, - "output_name": "output" - } - }, - "inputs": [], - "name": "listfiltering", - "outputs": [ - { - "name": "output_file", - "type": "pileup" - } - ], - "position": { - "left": 1084, - "top": 677.5 - }, - "post_job_actions": {}, - "tool_errors": null, - "tool_id": "listfiltering", - "tool_state": "{\"__page__\": 0, \"__rerun_remap_job_id__\": null, \"nb_col\": \"\\\"2\\\"\", \"input_listB\": \"null\", \"input_listA\": \"null\", \"chromInfo\": \"\\\"/home/galaxy/galaxy-dist/tool-data/shared/ucsc/chrom/?.len\\\"\"}", - "tool_version": "0.01", - "type": "tool", - "user_outputs": [] - }, - "8": { - "annotation": "", - "id": 8, - "input_connections": { - "input_assembly_file": { - "id": 1, - "output_name": "output" - }, - "input_variant_file": { - "id": 7, - "output_name": "output_file" - } - }, - "inputs": [], - "name": "extractseq", - "outputs": [ - { - "name": "output_file", - "type": "fasta" - } - ], - "position": { - "left": 1412, - "top": 677.5 - }, - "post_job_actions": {}, - "tool_errors": null, - "tool_id": "extractseq", - "tool_state": "{\"__page__\": 0, \"input_variant_file\": \"null\", \"__rerun_remap_job_id__\": null, \"input_assembly_file\": \"null\", \"window_length\": \"\\\"50\\\"\", \"chromInfo\": \"\\\"/home/galaxy/galaxy-dist/tool-data/shared/ucsc/chrom/?.len\\\"\"}", - "tool_version": "0.03", - "type": "tool", - "user_outputs": [] - }, - "9": { - "annotation": "", - "id": 9, - "input_connections": { - "db_opts|histdb": { - "id": 6, - "output_name": "outfile" - }, - "query": { - "id": 8, - "output_name": "output_file" - } - }, - "inputs": [], - "name": "NCBI BLAST+ blastn", - "outputs": [ - { - "name": "output1", - "type": "tabular" - } - ], - "position": { - "left": 1740, - "top": 677.5 - }, - "post_job_actions": {}, - "tool_errors": null, - "tool_id": "toolshed.g2.bx.psu.edu/repos/devteam/ncbi_blast_plus/ncbi_blastn_wrapper/0.1.00", - "tool_state": "{\"evalue_cutoff\": \"\\\"0.001\\\"\", \"__page__\": 0, \"adv_opts\": \"{\\\"identity_cutoff\\\": \\\"0.0\\\", \\\"adv_opts_selector\\\": \\\"advanced\\\", \\\"ungapped\\\": \\\"False\\\", \\\"filter_query\\\": \\\"False\\\", \\\"word_size\\\": \\\"0\\\", \\\"__current_case__\\\": 1, \\\"parse_deflines\\\": \\\"False\\\", \\\"strand\\\": \\\"-strand both\\\", \\\"max_hits\\\": \\\"0\\\"}\", \"__rerun_remap_job_id__\": null, \"db_opts\": \"{\\\"db_opts_selector\\\": \\\"histdb\\\", \\\"subject\\\": \\\"\\\", \\\"histdb\\\": null, \\\"__current_case__\\\": 1, \\\"database\\\": \\\"\\\"}\", \"query\": \"null\", \"blast_type\": \"\\\"megablast\\\"\", \"output\": \"{\\\"out_format\\\": \\\"5\\\", \\\"__current_case__\\\": 3}\", \"chromInfo\": \"\\\"/home/galaxy/galaxy-dist/tool-data/shared/ucsc/chrom/?.len\\\"\"}", - "tool_version": "0.1.00", - "type": "tool", - "user_outputs": [] - }, - "10": { - "annotation": "", - "id": 10, - "input_connections": { - "input_blastxml_file": { - "id": 9, - "output_name": "output1" - }, - "input_variant_file": { - "id": 7, - "output_name": "output_file" - } - }, - "inputs": [], - "name": "mpileupfilteronblastxml", - "outputs": [ - { - "name": "output_file", - "type": "pileup" - } - ], - "position": { - "left": 2042, - "top": 677.5 - }, - "post_job_actions": {}, - "tool_errors": null, - "tool_id": "mpileupfilteronblastxml", - "tool_state": "{\"__page__\": 0, \"input_variant_file\": \"null\", \"__rerun_remap_job_id__\": null, \"input_blastxml_file\": \"null\", \"window_length\": \"\\\"50\\\"\", \"nb_mismatch_max\": \"\\\"3\\\"\", \"chromInfo\": \"\\\"/home/galaxy/galaxy-dist/tool-data/shared/ucsc/chrom/?.len\\\"\"}", - "tool_version": "0.03", - "type": "tool", - "user_outputs": [] - }, - "11": { - "annotation": "", - "id": 11, - "input_connections": { - "input_blast_filtered_file": { - "id": 10, - "output_name": "output_file" - }, - "input_list_filtered_file": { - "id": 7, - "output_name": "output_file" - }, - "input_mpileup_filtered_file": { - "id": 5, - "output_name": "output_file" - }, - "input_mpileup_row_file": { - "id": 3, - "output_name": "output_mpileup" - } - }, - "inputs": [], - "name": "rapsodyn_stats_filtering", - "outputs": [ - { - "name": "output_file", - "type": "txt" - } - ], - "position": { - "left": 1683.5, - "top": 233 - }, - "post_job_actions": {}, - "tool_errors": null, - "tool_id": "rapsodyn_stats_filtering", - "tool_state": "{\"__page__\": 0, \"__rerun_remap_job_id__\": null, \"input_mpileup_filtered_file\": \"null\", \"input_mpileup_row_file\": \"null\", \"input_list_filtered_file\": \"null\", \"input_blast_filtered_file\": \"null\"}", - "tool_version": "0.01", - "type": "tool", - "user_outputs": [] - } - } -} \ No newline at end of file diff -r edddaa8ab855 -r e8e6b962c1f2 rapsodyn/Galaxy-Workflow-Rapsodyn_v1.34_-_Preparation___Mapping.ga --- a/rapsodyn/Galaxy-Workflow-Rapsodyn_v1.34_-_Preparation___Mapping.ga Thu Aug 21 08:39:15 2014 -0400 +++ /dev/null Thu Jan 01 00:00:00 1970 +0000 @@ -1,393 +0,0 @@ -{ - "a_galaxy_workflow": "true", - "annotation": "", - "format-version": "0.1", - "name": "Rapsodyn v1.34 - Preparation & Mapping", - "steps": { - "0": { - "annotation": "", - "id": 0, - "input_connections": {}, - "inputs": [ - { - "description": "", - "name": "READ1" - } - ], - "name": "Input dataset", - "outputs": [], - "position": { - "left": 440.2833251953125, - "top": 200 - }, - "tool_errors": null, - "tool_id": null, - "tool_state": "{\"name\": \"READ1\"}", - "tool_version": null, - "type": "data_input", - "user_outputs": [] - }, - "1": { - "annotation": "", - "id": 1, - "input_connections": {}, - "inputs": [ - { - "description": "", - "name": "READ2" - } - ], - "name": "Input dataset", - "outputs": [], - "position": { - "left": 440.2833251953125, - "top": 284.99998474121094 - }, - "tool_errors": null, - "tool_id": null, - "tool_state": "{\"name\": \"READ2\"}", - "tool_version": null, - "type": "data_input", - "user_outputs": [] - }, - "2": { - "annotation": "", - "id": 2, - "input_connections": {}, - "inputs": [ - { - "description": "", - "name": "ASSEMBLY" - } - ], - "name": "Input dataset", - "outputs": [], - "position": { - "left": 440.2833251953125, - "top": 369.99998474121094 - }, - "tool_errors": null, - "tool_id": null, - "tool_state": "{\"name\": \"ASSEMBLY\"}", - "tool_version": null, - "type": "data_input", - "user_outputs": [] - }, - "3": { - "annotation": "", - "id": 3, - "input_connections": { - "input_file": { - "id": 0, - "output_name": "output" - } - }, - "inputs": [], - "name": "FASTQ Groomer", - "outputs": [ - { - "name": "output_file", - "type": "fastqsanger" - } - ], - "position": { - "left": 668.2833251953125, - "top": 200 - }, - "post_job_actions": {}, - "tool_errors": null, - "tool_id": "toolshed.g2.bx.psu.edu/repos/devteam/fastq_groomer/fastq_groomer/1.0.4", - "tool_state": "{\"__page__\": 0, \"input_file\": \"null\", \"input_type\": \"\\\"illumina\\\"\", \"__rerun_remap_job_id__\": null, \"options_type\": \"{\\\"options_type_selector\\\": \\\"basic\\\", \\\"__current_case__\\\": 0}\", \"chromInfo\": \"\\\"/home/galaxy/galaxy-dist/tool-data/shared/ucsc/chrom/?.len\\\"\"}", - "tool_version": "1.0.4", - "type": "tool", - "user_outputs": [] - }, - "4": { - "annotation": "", - "id": 4, - "input_connections": { - "input_file": { - "id": 1, - "output_name": "output" - } - }, - "inputs": [], - "name": "FASTQ Groomer", - "outputs": [ - { - "name": "output_file", - "type": "fastqsanger" - } - ], - "position": { - "left": 668.2833251953125, - "top": 330.99998474121094 - }, - "post_job_actions": {}, - "tool_errors": null, - "tool_id": "toolshed.g2.bx.psu.edu/repos/devteam/fastq_groomer/fastq_groomer/1.0.4", - "tool_state": "{\"__page__\": 0, \"input_file\": \"null\", \"input_type\": \"\\\"illumina\\\"\", \"__rerun_remap_job_id__\": null, \"options_type\": \"{\\\"options_type_selector\\\": \\\"basic\\\", \\\"__current_case__\\\": 0}\", \"chromInfo\": \"\\\"/home/galaxy/galaxy-dist/tool-data/shared/ucsc/chrom/?.len\\\"\"}", - "tool_version": "1.0.4", - "type": "tool", - "user_outputs": [] - }, - "5": { - "annotation": "", - "id": 5, - "input_connections": { - "readtype|input_paired1": { - "id": 3, - "output_name": "output_file" - }, - "readtype|input_paired2": { - "id": 4, - "output_name": "output_file" - } - }, - "inputs": [], - "name": "Sickle", - "outputs": [ - { - "name": "output_single", - "type": "input" - }, - { - "name": "output_combo", - "type": "input" - }, - { - "name": "output_combo_single", - "type": "input" - }, - { - "name": "output_paired1", - "type": "fastq" - }, - { - "name": "output_paired2", - "type": "fastq" - }, - { - "name": "output_paired_single", - "type": "fastq" - } - ], - "position": { - "left": 996.2833251953125, - "top": 200 - }, - "post_job_actions": {}, - "tool_errors": null, - "tool_id": "toolshed.g2.bx.psu.edu/repos/nikhil-joshi/sickle/sickle/1.33", - "tool_state": "{\"__page__\": 0, \"length_threshold\": \"\\\"30\\\"\", \"no_five_prime\": \"\\\"False\\\"\", \"__rerun_remap_job_id__\": null, \"readtype\": \"{\\\"single_or_paired\\\": \\\"pe_sep\\\", \\\"input_paired1\\\": null, \\\"input_paired2\\\": null, \\\"__current_case__\\\": 2}\", \"trunc_n\": \"\\\"False\\\"\", \"qual_threshold\": \"\\\"30\\\"\"}", - "tool_version": "1.33", - "type": "tool", - "user_outputs": [] - }, - "6": { - "annotation": "", - "id": 6, - "input_connections": { - "readtype|input_paired1": { - "id": 5, - "output_name": "output_paired1" - }, - "readtype|input_paired2": { - "id": 5, - "output_name": "output_paired2" - } - }, - "inputs": [], - "name": "Sickle", - "outputs": [ - { - "name": "output_single", - "type": "input" - }, - { - "name": "output_combo", - "type": "input" - }, - { - "name": "output_combo_single", - "type": "input" - }, - { - "name": "output_paired1", - "type": "fastq" - }, - { - "name": "output_paired2", - "type": "fastq" - }, - { - "name": "output_paired_single", - "type": "fastq" - } - ], - "position": { - "left": 1324.2833251953125, - "top": 200 - }, - "post_job_actions": {}, - "tool_errors": null, - "tool_id": "toolshed.g2.bx.psu.edu/repos/nikhil-joshi/sickle/sickle/1.33", - "tool_state": "{\"__page__\": 0, \"length_threshold\": \"\\\"30\\\"\", \"no_five_prime\": \"\\\"False\\\"\", \"__rerun_remap_job_id__\": null, \"readtype\": \"{\\\"single_or_paired\\\": \\\"pe_sep\\\", \\\"input_paired1\\\": null, \\\"input_paired2\\\": null, \\\"__current_case__\\\": 2}\", \"trunc_n\": \"\\\"True\\\"\", \"qual_threshold\": \"\\\"30\\\"\"}", - "tool_version": "1.33", - "type": "tool", - "user_outputs": [] - }, - "7": { - "annotation": "", - "id": 7, - "input_connections": { - "genomeSource|ownFile": { - "id": 2, - "output_name": "output" - }, - "paired|input1": { - "id": 6, - "output_name": "output_paired1" - }, - "paired|input2": { - "id": 6, - "output_name": "output_paired2" - } - }, - "inputs": [], - "name": "Map with BWA for Illumina", - "outputs": [ - { - "name": "output", - "type": "sam" - } - ], - "position": { - "left": 1648.2833251953125, - "top": 200 - }, - "post_job_actions": {}, - "tool_errors": null, - "tool_id": "toolshed.g2.bx.psu.edu/repos/devteam/bwa_wrappers/bwa_wrapper/1.2.3", - "tool_state": "{\"genomeSource\": \"{\\\"refGenomeSource\\\": \\\"history\\\", \\\"ownFile\\\": null, \\\"__current_case__\\\": 1}\", \"__page__\": 0, \"__rerun_remap_job_id__\": null, \"paired\": \"{\\\"input2\\\": null, \\\"sPaired\\\": \\\"paired\\\", \\\"input1\\\": null, \\\"__current_case__\\\": 1}\", \"params\": \"{\\\"__current_case__\\\": 0, \\\"source_select\\\": \\\"pre_set\\\"}\", \"suppressHeader\": \"\\\"False\\\"\", \"chromInfo\": \"\\\"/home/galaxy/galaxy-dist/tool-data/shared/ucsc/chrom/?.len\\\"\"}", - "tool_version": "1.2.3", - "type": "tool", - "user_outputs": [] - }, - "8": { - "annotation": "", - "id": 8, - "input_connections": { - "input_sam_file": { - "id": 7, - "output_name": "output" - } - }, - "inputs": [], - "name": "filtersam_mapped_and_unique", - "outputs": [ - { - "name": "output_file", - "type": "sam" - } - ], - "position": { - "left": 1912.283447265625, - "top": 200 - }, - "post_job_actions": {}, - "tool_errors": null, - "tool_id": "filtersam_mapped_and_unique", - "tool_state": "{\"__page__\": 0, \"__rerun_remap_job_id__\": null, \"input_sam_file\": \"null\"}", - "tool_version": "0.01", - "type": "tool", - "user_outputs": [] - }, - "9": { - "annotation": "", - "id": 9, - "input_connections": { - "input_read1_row_file": { - "id": 0, - "output_name": "output" - }, - "input_read1_trimmed_file": { - "id": 5, - "output_name": "output_paired1" - }, - "input_read1_trimmednoN_file": { - "id": 6, - "output_name": "output_paired1" - }, - "input_read2_row_file": { - "id": 1, - "output_name": "output" - }, - "input_read2_trimmed_file": { - "id": 5, - "output_name": "output_paired2" - }, - "input_read2_trimmednoN_file": { - "id": 6, - "output_name": "output_paired2" - }, - "input_sam_filtered_file": { - "id": 8, - "output_name": "output_file" - }, - "input_sam_row_file": { - "id": 7, - "output_name": "output" - } - }, - "inputs": [], - "name": "rapsodyn_stats_mapping", - "outputs": [ - { - "name": "output_file", - "type": "txt" - } - ], - "position": { - "left": 2240.283447265625, - "top": 200 - }, - "post_job_actions": {}, - "tool_errors": null, - "tool_id": "rapsodyn_stats_mapping", - "tool_state": "{\"input_read1_trimmed_file\": \"null\", \"input_read2_trimmednoN_file\": \"null\", \"input_read2_trimmed_file\": \"null\", \"__page__\": 0, \"__rerun_remap_job_id__\": null, \"input_read2_row_file\": \"null\", \"input_read1_trimmednoN_file\": \"null\", \"input_sam_filtered_file\": \"null\", \"input_read1_row_file\": \"null\", \"input_sam_row_file\": \"null\"}", - "tool_version": "0.01", - "type": "tool", - "user_outputs": [] - }, - "10": { - "annotation": "", - "id": 10, - "input_connections": { - "source|input1": { - "id": 8, - "output_name": "output_file" - } - }, - "inputs": [], - "name": "SAM-to-BAM", - "outputs": [ - { - "name": "output1", - "type": "bam" - } - ], - "position": { - "left": 2240.283447265625, - "top": 640.9999847412109 - }, - "post_job_actions": {}, - "tool_errors": null, - "tool_id": "toolshed.g2.bx.psu.edu/repos/devteam/sam_to_bam/sam_to_bam/1.1.4", - "tool_state": "{\"source\": \"{\\\"index_source\\\": \\\"cached\\\", \\\"index\\\": {\\\"__class__\\\": \\\"UnvalidatedValue\\\", \\\"value\\\": null}, \\\"input1\\\": null, \\\"__current_case__\\\": 0}\", \"__rerun_remap_job_id__\": null, \"__page__\": 0}", - "tool_version": "1.1.4", - "type": "tool", - "user_outputs": [] - } - } -} \ No newline at end of file diff -r edddaa8ab855 -r e8e6b962c1f2 rapsodyn/Galaxy-Workflow-rapsosnp_v1.13.ga --- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/rapsodyn/Galaxy-Workflow-rapsosnp_v1.13.ga Fri Sep 05 06:12:10 2014 -0400 @@ -0,0 +1,525 @@ +{ + "a_galaxy_workflow": "true", + "annotation": "", + "format-version": "0.1", + "name": "rapsosnp v1.13", + "steps": { + "0": { + "annotation": "", + "id": 0, + "input_connections": {}, + "inputs": [ + { + "description": "", + "name": "ASSEMBLY" + } + ], + "name": "Input dataset", + "outputs": [], + "position": { + "left": 200, + "top": 297 + }, + "tool_errors": null, + "tool_id": null, + "tool_state": "{\"name\": \"ASSEMBLY\"}", + "tool_version": null, + "type": "data_input", + "user_outputs": [] + }, + "1": { + "annotation": "", + "id": 1, + "input_connections": {}, + "inputs": [ + { + "description": "", + "name": "READ1" + } + ], + "name": "Input dataset", + "outputs": [], + "position": { + "left": 200, + "top": 382 + }, + "tool_errors": null, + "tool_id": null, + "tool_state": "{\"name\": \"READ1\"}", + "tool_version": null, + "type": "data_input", + "user_outputs": [] + }, + "2": { + "annotation": "", + "id": 2, + "input_connections": {}, + "inputs": [ + { + "description": "", + "name": "READ2" + } + ], + "name": "Input dataset", + "outputs": [], + "position": { + "left": 208, + "top": 472 + }, + "tool_errors": null, + "tool_id": null, + "tool_state": "{\"name\": \"READ2\"}", + "tool_version": null, + "type": "data_input", + "user_outputs": [] + }, + "3": { + "annotation": "", + "id": 3, + "input_connections": {}, + "inputs": [ + { + "description": "", + "name": "DUBIOUS" + } + ], + "name": "Input dataset", + "outputs": [], + "position": { + "left": 200, + "top": 552 + }, + "tool_errors": null, + "tool_id": null, + "tool_state": "{\"name\": \"DUBIOUS\"}", + "tool_version": null, + "type": "data_input", + "user_outputs": [] + }, + "4": { + "annotation": "", + "id": 4, + "input_connections": { + "input_fasta": { + "id": 0, + "output_name": "output" + } + }, + "inputs": [], + "name": "fastaGroomerForMakeBlastdb", + "outputs": [ + { + "name": "output_fasta", + "type": "fasta" + } + ], + "position": { + "left": 428, + "top": 297 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "toolshed.g2.bx.psu.edu/repos/mcharles/genephys/fastaGroomerForMakeBlastdb/0.01", + "tool_state": "{\"input_fasta\": \"null\", \"__rerun_remap_job_id__\": null, \"__page__\": 0}", + "tool_version": "0.01", + "type": "tool", + "user_outputs": [] + }, + "5": { + "annotation": "", + "id": 5, + "input_connections": { + "input_read1_file": { + "id": 1, + "output_name": "output" + }, + "input_read2_file": { + "id": 2, + "output_name": "output" + } + }, + "inputs": [], + "name": "PrepareFastqLight", + "outputs": [ + { + "name": "output_read1_file", + "type": "fastqsanger" + }, + { + "name": "output_read2_file", + "type": "fastqsanger" + } + ], + "position": { + "left": 428, + "top": 428 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "PrepareFastqLight", + "tool_state": "{\"__page__\": 0, \"quality_type\": \"\\\"illumina\\\"\", \"min_length\": \"\\\"30\\\"\", \"__rerun_remap_job_id__\": null, \"min_quality\": \"\\\"30\\\"\", \"input_read1_file\": \"null\", \"input_read2_file\": \"null\"}", + "tool_version": "0.01", + "type": "tool", + "user_outputs": [] + }, + "6": { + "annotation": "", + "id": 6, + "input_connections": { + "input_file": { + "id": 4, + "output_name": "output_fasta" + } + }, + "inputs": [], + "name": "NCBI BLAST+ makeblastdb", + "outputs": [ + { + "name": "outfile", + "type": "data" + } + ], + "position": { + "left": 756, + "top": 297 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "toolshed.g2.bx.psu.edu/repos/devteam/ncbi_blast_plus/ncbi_makeblastdb/0.1.00", + "tool_state": "{\"__page__\": 0, \"mask_data_file\": \"null\", \"input_file\": \"null\", \"dbtype\": \"\\\"nucl\\\"\", \"__rerun_remap_job_id__\": null, \"hash_index\": \"\\\"True\\\"\", \"tax\": \"{\\\"taxselect\\\": \\\"\\\", \\\"__current_case__\\\": 0}\", \"title\": \"\\\"\\\"\", \"parse_seqids\": \"\\\"False\\\"\"}", + "tool_version": "0.1.00", + "type": "tool", + "user_outputs": [] + }, + "7": { + "annotation": "", + "id": 7, + "input_connections": { + "genomeSource|ownFile": { + "id": 0, + "output_name": "output" + }, + "paired|input1": { + "id": 5, + "output_name": "output_read1_file" + }, + "paired|input2": { + "id": 5, + "output_name": "output_read2_file" + } + }, + "inputs": [], + "name": "Map with BWA for Illumina", + "outputs": [ + { + "name": "output", + "type": "sam" + } + ], + "position": { + "left": 756, + "top": 455 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "toolshed.g2.bx.psu.edu/repos/devteam/bwa_wrappers/bwa_wrapper/1.2.3", + "tool_state": "{\"genomeSource\": \"{\\\"refGenomeSource\\\": \\\"history\\\", \\\"ownFile\\\": null, \\\"__current_case__\\\": 1}\", \"__page__\": 0, \"__rerun_remap_job_id__\": null, \"paired\": \"{\\\"input2\\\": null, \\\"sPaired\\\": \\\"paired\\\", \\\"input1\\\": null, \\\"__current_case__\\\": 1}\", \"params\": \"{\\\"__current_case__\\\": 0, \\\"source_select\\\": \\\"pre_set\\\"}\", \"suppressHeader\": \"\\\"False\\\"\", \"chromInfo\": \"\\\"/home/galaxy/galaxy/galaxy-dist/tool-data/shared/ucsc/chrom/?.len\\\"\"}", + "tool_version": "1.2.3", + "type": "tool", + "user_outputs": [] + }, + "8": { + "annotation": "", + "id": 8, + "input_connections": { + "input_sam_file": { + "id": 7, + "output_name": "output" + } + }, + "inputs": [], + "name": "filtersam_mapped_and_unique", + "outputs": [ + { + "name": "output_file", + "type": "sam" + } + ], + "position": { + "left": 1084, + "top": 297 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "toolshed.g2.bx.psu.edu/repos/mcharles/rapsodyn/filtersam_mapped_and_unique/0.01", + "tool_state": "{\"__page__\": 0, \"__rerun_remap_job_id__\": null, \"chromInfo\": \"\\\"/home/galaxy/galaxy/galaxy-dist/tool-data/shared/ucsc/chrom/?.len\\\"\", \"input_sam_file\": \"null\"}", + "tool_version": "0.01", + "type": "tool", + "user_outputs": [] + }, + "9": { + "annotation": "", + "id": 9, + "input_connections": { + "source|input1": { + "id": 8, + "output_name": "output_file" + }, + "source|ref_file": { + "id": 0, + "output_name": "output" + } + }, + "inputs": [], + "name": "SAM-to-BAM", + "outputs": [ + { + "name": "output1", + "type": "bam" + } + ], + "position": { + "left": 1412, + "top": 297 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "toolshed.g2.bx.psu.edu/repos/devteam/sam_to_bam/sam_to_bam/1.1.4", + "tool_state": "{\"source\": \"{\\\"index_source\\\": \\\"history\\\", \\\"ref_file\\\": null, \\\"input1\\\": null, \\\"__current_case__\\\": 1}\", \"__rerun_remap_job_id__\": null, \"chromInfo\": \"\\\"/home/galaxy/galaxy/galaxy-dist/tool-data/shared/ucsc/chrom/?.len\\\"\", \"__page__\": 0}", + "tool_version": "1.1.4", + "type": "tool", + "user_outputs": [] + }, + "10": { + "annotation": "", + "id": 10, + "input_connections": { + "reference_source|input_bams_0|input_bam": { + "id": 9, + "output_name": "output1" + }, + "reference_source|ref_file": { + "id": 0, + "output_name": "output" + } + }, + "inputs": [], + "name": "MPileup", + "outputs": [ + { + "name": "output_mpileup", + "type": "pileup" + }, + { + "name": "output_log", + "type": "txt" + } + ], + "position": { + "left": 1640, + "top": 297 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "toolshed.g2.bx.psu.edu/repos/devteam/samtools_mpileup/samtools_mpileup/0.0.3", + "tool_state": "{\"__page__\": 0, \"genotype_likelihood_computation_type\": \"{\\\"genotype_likelihood_computation_type_selector\\\": \\\"do_not_perform_genotype_likelihood_computation\\\", \\\"__current_case__\\\": 1}\", \"__rerun_remap_job_id__\": null, \"advanced_options\": \"{\\\"max_reads_per_bam\\\": \\\"250\\\", \\\"advanced_options_selector\\\": \\\"advanced\\\", \\\"extended_BAQ_computation\\\": \\\"False\\\", \\\"region_string\\\": \\\"\\\", \\\"output_per_sample_strand_bias_p_value\\\": \\\"False\\\", \\\"minimum_base_quality\\\": \\\"30\\\", \\\"disable_probabilistic_realignment\\\": \\\"False\\\", \\\"skip_anomalous_read_pairs\\\": \\\"False\\\", \\\"minimum_mapping_quality\\\": \\\"0\\\", \\\"output_per_sample_read_depth\\\": \\\"False\\\", \\\"__current_case__\\\": 0, \\\"position_list\\\": null, \\\"coefficient_for_downgrading\\\": \\\"0\\\"}\", \"reference_source\": \"{\\\"ref_file\\\": null, \\\"reference_source_selector\\\": \\\"history\\\", \\\"input_bams\\\": [{\\\"__index__\\\": 0, \\\"input_bam\\\": null}], \\\"__current_case__\\\": 1}\", \"chromInfo\": \"\\\"/home/galaxy/galaxy/galaxy-dist/tool-data/shared/ucsc/chrom/?.len\\\"\"}", + "tool_version": "0.0.3", + "type": "tool", + "user_outputs": [] + }, + "11": { + "annotation": "", + "id": 11, + "input_connections": { + "input_file": { + "id": 10, + "output_name": "output_mpileup" + } + }, + "inputs": [], + "name": "PileupVariant", + "outputs": [ + { + "name": "output_file", + "type": "pileup" + } + ], + "position": { + "left": 1966, + "top": 297 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "PileupVariant", + "tool_state": "{\"__page__\": 0, \"__rerun_remap_job_id__\": null, \"input_file\": \"null\"}", + "tool_version": "0.01", + "type": "tool", + "user_outputs": [] + }, + "12": { + "annotation": "", + "id": 12, + "input_connections": { + "input_listA": { + "id": 11, + "output_name": "output_file" + }, + "input_listB": { + "id": 3, + "output_name": "output" + } + }, + "inputs": [], + "name": "listfiltering", + "outputs": [ + { + "name": "output_file", + "type": "pileup" + } + ], + "position": { + "left": 2294, + "top": 297 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "toolshed.g2.bx.psu.edu/repos/mcharles/rapsodyn/listfiltering/0.01", + "tool_state": "{\"__page__\": 0, \"__rerun_remap_job_id__\": null, \"input_listB\": \"null\", \"nb_col\": \"\\\"2\\\"\", \"input_listA\": \"null\"}", + "tool_version": "0.01", + "type": "tool", + "user_outputs": [] + }, + "13": { + "annotation": "", + "id": 13, + "input_connections": { + "input_assembly_file": { + "id": 0, + "output_name": "output" + }, + "input_variant_file": { + "id": 12, + "output_name": "output_file" + } + }, + "inputs": [], + "name": "extractseq", + "outputs": [ + { + "name": "output_file", + "type": "fasta" + } + ], + "position": { + "left": 2622, + "top": 297 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "toolshed.g2.bx.psu.edu/repos/mcharles/rapsodyn/extractseq/0.03", + "tool_state": "{\"__page__\": 0, \"input_variant_file\": \"null\", \"__rerun_remap_job_id__\": null, \"window_length\": \"\\\"50\\\"\", \"input_assembly_file\": \"null\"}", + "tool_version": "0.03", + "type": "tool", + "user_outputs": [] + }, + "14": { + "annotation": "", + "id": 14, + "input_connections": { + "db_opts|histdb": { + "id": 6, + "output_name": "outfile" + }, + "query": { + "id": 13, + "output_name": "output_file" + } + }, + "inputs": [], + "name": "NCBI BLAST+ blastn", + "outputs": [ + { + "name": "output1", + "type": "tabular" + } + ], + "position": { + "left": 2950, + "top": 297 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "toolshed.g2.bx.psu.edu/repos/devteam/ncbi_blast_plus/ncbi_blastn_wrapper/0.1.00", + "tool_state": "{\"evalue_cutoff\": \"\\\"0.001\\\"\", \"__page__\": 0, \"adv_opts\": \"{\\\"identity_cutoff\\\": \\\"0.0\\\", \\\"adv_opts_selector\\\": \\\"advanced\\\", \\\"ungapped\\\": \\\"False\\\", \\\"filter_query\\\": \\\"False\\\", \\\"word_size\\\": \\\"0\\\", \\\"__current_case__\\\": 1, \\\"parse_deflines\\\": \\\"False\\\", \\\"strand\\\": \\\"-strand both\\\", \\\"max_hits\\\": \\\"0\\\"}\", \"__rerun_remap_job_id__\": null, \"blast_type\": \"\\\"megablast\\\"\", \"db_opts\": \"{\\\"db_opts_selector\\\": \\\"histdb\\\", \\\"subject\\\": \\\"\\\", \\\"histdb\\\": null, \\\"__current_case__\\\": 1, \\\"database\\\": \\\"\\\"}\", \"output\": \"{\\\"out_format\\\": \\\"5\\\", \\\"__current_case__\\\": 3}\", \"query\": \"null\"}", + "tool_version": "0.1.00", + "type": "tool", + "user_outputs": [] + }, + "15": { + "annotation": "", + "id": 15, + "input_connections": { + "input_blastxml_file": { + "id": 14, + "output_name": "output1" + }, + "input_variant_file": { + "id": 12, + "output_name": "output_file" + } + }, + "inputs": [], + "name": "mpileupfilteronblastxml", + "outputs": [ + { + "name": "output_file", + "type": "pileup" + } + ], + "position": { + "left": 3252, + "top": 297 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "toolshed.g2.bx.psu.edu/repos/mcharles/rapsodyn/mpileupfilteronblastxml/0.03", + "tool_state": "{\"input_variant_file\": \"null\", \"__page__\": 0, \"__rerun_remap_job_id__\": null, \"input_blastxml_file\": \"null\", \"window_length\": \"\\\"50\\\"\", \"nb_mismatch_max\": \"\\\"3\\\"\"}", + "tool_version": "0.03", + "type": "tool", + "user_outputs": [] + }, + "16": { + "annotation": "", + "id": 16, + "input_connections": { + "input_file": { + "id": 15, + "output_name": "output_file" + } + }, + "inputs": [], + "name": "mpileupfilterandstat", + "outputs": [ + { + "name": "output_file", + "type": "pileup" + }, + { + "name": "log_file", + "type": "txt" + } + ], + "position": { + "left": 3607, + "top": 310.5 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "mpileupfilterandstat", + "tool_state": "{\"__page__\": 0, \"input_file\": \"null\", \"min_frequency\": \"\\\"0.8\\\"\", \"do_stat\": \"\\\"YES\\\"\", \"min_depth\": \"\\\"2\\\"\", \"__rerun_remap_job_id__\": null, \"min_forward_and_reverse\": \"\\\"0\\\"\", \"max_depth\": \"\\\"100\\\"\", \"min_distance\": \"\\\"0\\\"\"}", + "tool_version": "0.07", + "type": "tool", + "user_outputs": [] + } + } +} \ No newline at end of file diff -r edddaa8ab855 -r e8e6b962c1f2 rapsodyn/Galaxy-Workflow-rapsosnp_v1.13_-_2x.ga --- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/rapsodyn/Galaxy-Workflow-rapsosnp_v1.13_-_2x.ga Fri Sep 05 06:12:10 2014 -0400 @@ -0,0 +1,929 @@ +{ + "a_galaxy_workflow": "true", + "annotation": "", + "format-version": "0.1", + "name": "rapsosnp v1.13 - 2x", + "steps": { + "0": { + "annotation": "", + "id": 0, + "input_connections": {}, + "inputs": [ + { + "description": "", + "name": "ASSEMBLY" + } + ], + "name": "Input dataset", + "outputs": [], + "position": { + "left": 200, + "top": 200 + }, + "tool_errors": null, + "tool_id": null, + "tool_state": "{\"name\": \"ASSEMBLY\"}", + "tool_version": null, + "type": "data_input", + "user_outputs": [] + }, + "1": { + "annotation": "", + "id": 1, + "input_connections": {}, + "inputs": [ + { + "description": "", + "name": "READ1" + } + ], + "name": "Input dataset", + "outputs": [], + "position": { + "left": 200, + "top": 285 + }, + "tool_errors": null, + "tool_id": null, + "tool_state": "{\"name\": \"READ1\"}", + "tool_version": null, + "type": "data_input", + "user_outputs": [] + }, + "2": { + "annotation": "", + "id": 2, + "input_connections": {}, + "inputs": [ + { + "description": "", + "name": "READ2" + } + ], + "name": "Input dataset", + "outputs": [], + "position": { + "left": 200, + "top": 370 + }, + "tool_errors": null, + "tool_id": null, + "tool_state": "{\"name\": \"READ2\"}", + "tool_version": null, + "type": "data_input", + "user_outputs": [] + }, + "3": { + "annotation": "", + "id": 3, + "input_connections": {}, + "inputs": [ + { + "description": "", + "name": "DUBIOUS" + } + ], + "name": "Input dataset", + "outputs": [], + "position": { + "left": 200, + "top": 455 + }, + "tool_errors": null, + "tool_id": null, + "tool_state": "{\"name\": \"DUBIOUS\"}", + "tool_version": null, + "type": "data_input", + "user_outputs": [] + }, + "4": { + "annotation": "", + "id": 4, + "input_connections": { + "input_fasta": { + "id": 0, + "output_name": "output" + } + }, + "inputs": [], + "name": "fastaGroomerForMakeBlastdb", + "outputs": [ + { + "name": "output_fasta", + "type": "fasta" + } + ], + "position": { + "left": 428, + "top": 200 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "toolshed.g2.bx.psu.edu/repos/mcharles/genephys/fastaGroomerForMakeBlastdb/0.01", + "tool_state": "{\"input_fasta\": \"null\", \"__rerun_remap_job_id__\": null, \"__page__\": 0}", + "tool_version": "0.01", + "type": "tool", + "user_outputs": [] + }, + "5": { + "annotation": "", + "id": 5, + "input_connections": { + "input_file": { + "id": 1, + "output_name": "output" + } + }, + "inputs": [], + "name": "SplitFile2x", + "outputs": [ + { + "name": "output_file1", + "type": "txt" + }, + { + "name": "output_file2", + "type": "txt" + } + ], + "position": { + "left": 428, + "top": 331 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "SplitFile2x", + "tool_state": "{\"out_format\": \"\\\"fastq\\\"\", \"line_number\": \"\\\"4\\\"\", \"__page__\": 0, \"__rerun_remap_job_id__\": null, \"input_file\": \"null\"}", + "tool_version": "0.01", + "type": "tool", + "user_outputs": [] + }, + "6": { + "annotation": "", + "id": 6, + "input_connections": { + "input_file": { + "id": 2, + "output_name": "output" + } + }, + "inputs": [], + "name": "SplitFile2x", + "outputs": [ + { + "name": "output_file1", + "type": "txt" + }, + { + "name": "output_file2", + "type": "txt" + } + ], + "position": { + "left": 428, + "top": 523 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "SplitFile2x", + "tool_state": "{\"out_format\": \"\\\"fastq\\\"\", \"line_number\": \"\\\"4\\\"\", \"__page__\": 0, \"__rerun_remap_job_id__\": null, \"input_file\": \"null\"}", + "tool_version": "0.01", + "type": "tool", + "user_outputs": [] + }, + "7": { + "annotation": "", + "id": 7, + "input_connections": { + "input_file": { + "id": 4, + "output_name": "output_fasta" + } + }, + "inputs": [], + "name": "NCBI BLAST+ makeblastdb", + "outputs": [ + { + "name": "outfile", + "type": "data" + } + ], + "position": { + "left": 756, + "top": 200 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "toolshed.g2.bx.psu.edu/repos/devteam/ncbi_blast_plus/ncbi_makeblastdb/0.1.00", + "tool_state": "{\"__page__\": 0, \"mask_data_file\": \"null\", \"input_file\": \"null\", \"dbtype\": \"\\\"nucl\\\"\", \"__rerun_remap_job_id__\": null, \"hash_index\": \"\\\"True\\\"\", \"tax\": \"{\\\"taxselect\\\": \\\"\\\", \\\"__current_case__\\\": 0}\", \"title\": \"\\\"\\\"\", \"parse_seqids\": \"\\\"False\\\"\"}", + "tool_version": "0.1.00", + "type": "tool", + "user_outputs": [] + }, + "8": { + "annotation": "", + "id": 8, + "input_connections": { + "input_read1_file": { + "id": 5, + "output_name": "output_file1" + }, + "input_read2_file": { + "id": 6, + "output_name": "output_file1" + } + }, + "inputs": [], + "name": "PrepareFastqLight", + "outputs": [ + { + "name": "output_read1_file", + "type": "fastqsanger" + }, + { + "name": "output_read2_file", + "type": "fastqsanger" + } + ], + "position": { + "left": 756, + "top": 358 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "PrepareFastqLight", + "tool_state": "{\"__page__\": 0, \"quality_type\": \"\\\"illumina\\\"\", \"min_length\": \"\\\"30\\\"\", \"__rerun_remap_job_id__\": null, \"min_quality\": \"\\\"30\\\"\", \"input_read1_file\": \"null\", \"input_read2_file\": \"null\"}", + "tool_version": "0.01", + "type": "tool", + "user_outputs": [] + }, + "9": { + "annotation": "", + "id": 9, + "input_connections": { + "input_read1_file": { + "id": 5, + "output_name": "output_file2" + }, + "input_read2_file": { + "id": 6, + "output_name": "output_file2" + } + }, + "inputs": [], + "name": "PrepareFastqLight", + "outputs": [ + { + "name": "output_read1_file", + "type": "fastqsanger" + }, + { + "name": "output_read2_file", + "type": "fastqsanger" + } + ], + "position": { + "left": 756, + "top": 560 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "PrepareFastqLight", + "tool_state": "{\"__page__\": 0, \"quality_type\": \"\\\"illumina\\\"\", \"min_length\": \"\\\"30\\\"\", \"__rerun_remap_job_id__\": null, \"min_quality\": \"\\\"30\\\"\", \"input_read1_file\": \"null\", \"input_read2_file\": \"null\"}", + "tool_version": "0.01", + "type": "tool", + "user_outputs": [] + }, + "10": { + "annotation": "", + "id": 10, + "input_connections": { + "genomeSource|ownFile": { + "id": 0, + "output_name": "output" + }, + "paired|input1": { + "id": 8, + "output_name": "output_read1_file" + }, + "paired|input2": { + "id": 8, + "output_name": "output_read2_file" + } + }, + "inputs": [], + "name": "Map with BWA for Illumina", + "outputs": [ + { + "name": "output", + "type": "sam" + } + ], + "position": { + "left": 1084, + "top": 200 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "toolshed.g2.bx.psu.edu/repos/devteam/bwa_wrappers/bwa_wrapper/1.2.3", + "tool_state": "{\"genomeSource\": \"{\\\"refGenomeSource\\\": \\\"history\\\", \\\"ownFile\\\": null, \\\"__current_case__\\\": 1}\", \"__page__\": 0, \"__rerun_remap_job_id__\": null, \"paired\": \"{\\\"input2\\\": null, \\\"sPaired\\\": \\\"paired\\\", \\\"input1\\\": null, \\\"__current_case__\\\": 1}\", \"params\": \"{\\\"__current_case__\\\": 0, \\\"source_select\\\": \\\"pre_set\\\"}\", \"suppressHeader\": \"\\\"False\\\"\", \"chromInfo\": \"\\\"/home/galaxy/galaxy/galaxy-dist/tool-data/shared/ucsc/chrom/?.len\\\"\"}", + "tool_version": "1.2.3", + "type": "tool", + "user_outputs": [] + }, + "11": { + "annotation": "", + "id": 11, + "input_connections": { + "genomeSource|ownFile": { + "id": 0, + "output_name": "output" + }, + "paired|input1": { + "id": 9, + "output_name": "output_read1_file" + }, + "paired|input2": { + "id": 9, + "output_name": "output_read2_file" + } + }, + "inputs": [], + "name": "Map with BWA for Illumina", + "outputs": [ + { + "name": "output", + "type": "sam" + } + ], + "position": { + "left": 1084, + "top": 385 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "toolshed.g2.bx.psu.edu/repos/devteam/bwa_wrappers/bwa_wrapper/1.2.3", + "tool_state": "{\"genomeSource\": \"{\\\"refGenomeSource\\\": \\\"history\\\", \\\"ownFile\\\": null, \\\"__current_case__\\\": 1}\", \"__page__\": 0, \"__rerun_remap_job_id__\": null, \"paired\": \"{\\\"input2\\\": null, \\\"sPaired\\\": \\\"paired\\\", \\\"input1\\\": null, \\\"__current_case__\\\": 1}\", \"params\": \"{\\\"__current_case__\\\": 0, \\\"source_select\\\": \\\"pre_set\\\"}\", \"suppressHeader\": \"\\\"False\\\"\", \"chromInfo\": \"\\\"/home/galaxy/galaxy/galaxy-dist/tool-data/shared/ucsc/chrom/?.len\\\"\"}", + "tool_version": "1.2.3", + "type": "tool", + "user_outputs": [] + }, + "12": { + "annotation": "", + "id": 12, + "input_connections": { + "input_sam_file": { + "id": 10, + "output_name": "output" + } + }, + "inputs": [], + "name": "filtersam_mapped_and_unique", + "outputs": [ + { + "name": "output_file", + "type": "sam" + } + ], + "position": { + "left": 1348, + "top": 200 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "toolshed.g2.bx.psu.edu/repos/mcharles/rapsodyn/filtersam_mapped_and_unique/0.01", + "tool_state": "{\"__page__\": 0, \"__rerun_remap_job_id__\": null, \"chromInfo\": \"\\\"/home/galaxy/galaxy/galaxy-dist/tool-data/shared/ucsc/chrom/?.len\\\"\", \"input_sam_file\": \"null\"}", + "tool_version": "0.01", + "type": "tool", + "user_outputs": [] + }, + "13": { + "annotation": "", + "id": 13, + "input_connections": { + "input_sam_file": { + "id": 11, + "output_name": "output" + } + }, + "inputs": [], + "name": "filtersam_mapped_and_unique", + "outputs": [ + { + "name": "output_file", + "type": "sam" + } + ], + "position": { + "left": 1348, + "top": 331 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "toolshed.g2.bx.psu.edu/repos/mcharles/rapsodyn/filtersam_mapped_and_unique/0.01", + "tool_state": "{\"__page__\": 0, \"__rerun_remap_job_id__\": null, \"chromInfo\": \"\\\"/home/galaxy/galaxy/galaxy-dist/tool-data/shared/ucsc/chrom/?.len\\\"\", \"input_sam_file\": \"null\"}", + "tool_version": "0.01", + "type": "tool", + "user_outputs": [] + }, + "14": { + "annotation": "", + "id": 14, + "input_connections": { + "source|input1": { + "id": 12, + "output_name": "output_file" + }, + "source|ref_file": { + "id": 0, + "output_name": "output" + } + }, + "inputs": [], + "name": "SAM-to-BAM", + "outputs": [ + { + "name": "output1", + "type": "bam" + } + ], + "position": { + "left": 1676, + "top": 200 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "toolshed.g2.bx.psu.edu/repos/devteam/sam_to_bam/sam_to_bam/1.1.4", + "tool_state": "{\"source\": \"{\\\"index_source\\\": \\\"history\\\", \\\"ref_file\\\": null, \\\"input1\\\": null, \\\"__current_case__\\\": 1}\", \"__rerun_remap_job_id__\": null, \"chromInfo\": \"\\\"/home/galaxy/galaxy/galaxy-dist/tool-data/shared/ucsc/chrom/?.len\\\"\", \"__page__\": 0}", + "tool_version": "1.1.4", + "type": "tool", + "user_outputs": [] + }, + "15": { + "annotation": "", + "id": 15, + "input_connections": { + "source|input1": { + "id": 13, + "output_name": "output_file" + }, + "source|ref_file": { + "id": 0, + "output_name": "output" + } + }, + "inputs": [], + "name": "SAM-to-BAM", + "outputs": [ + { + "name": "output1", + "type": "bam" + } + ], + "position": { + "left": 1676, + "top": 341 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "toolshed.g2.bx.psu.edu/repos/devteam/sam_to_bam/sam_to_bam/1.1.4", + "tool_state": "{\"source\": \"{\\\"index_source\\\": \\\"history\\\", \\\"ref_file\\\": null, \\\"input1\\\": null, \\\"__current_case__\\\": 1}\", \"__rerun_remap_job_id__\": null, \"chromInfo\": \"\\\"/home/galaxy/galaxy/galaxy-dist/tool-data/shared/ucsc/chrom/?.len\\\"\", \"__page__\": 0}", + "tool_version": "1.1.4", + "type": "tool", + "user_outputs": [] + }, + "16": { + "annotation": "", + "id": 16, + "input_connections": { + "input1": { + "id": 14, + "output_name": "output1" + }, + "input2": { + "id": 15, + "output_name": "output1" + } + }, + "inputs": [], + "name": "Merge BAM Files", + "outputs": [ + { + "name": "output1", + "type": "bam" + }, + { + "name": "outlog", + "type": "txt" + } + ], + "position": { + "left": 1904, + "top": 200 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "toolshed.g2.bx.psu.edu/repos/devteam/sam_merge/sam_merge2/1.1.2", + "tool_state": "{\"input2\": \"null\", \"inputs\": \"[]\", \"input1\": \"null\", \"title\": \"\\\"merge bam\\\"\", \"__page__\": 0, \"__rerun_remap_job_id__\": null, \"mergeSD\": \"\\\"True\\\"\", \"chromInfo\": \"\\\"/home/galaxy/galaxy/galaxy-dist/tool-data/shared/ucsc/chrom/?.len\\\"\"}", + "tool_version": "1.1.2", + "type": "tool", + "user_outputs": [] + }, + "17": { + "annotation": "", + "id": 17, + "input_connections": { + "reference_source|input_bams_0|input_bam": { + "id": 16, + "output_name": "output1" + }, + "reference_source|ref_file": { + "id": 0, + "output_name": "output" + } + }, + "inputs": [], + "name": "MPileup", + "outputs": [ + { + "name": "output_mpileup", + "type": "pileup" + }, + { + "name": "output_log", + "type": "txt" + } + ], + "position": { + "left": 2132, + "top": 200 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "toolshed.g2.bx.psu.edu/repos/devteam/samtools_mpileup/samtools_mpileup/0.0.3", + "tool_state": "{\"__page__\": 0, \"genotype_likelihood_computation_type\": \"{\\\"genotype_likelihood_computation_type_selector\\\": \\\"do_not_perform_genotype_likelihood_computation\\\", \\\"__current_case__\\\": 1}\", \"__rerun_remap_job_id__\": null, \"advanced_options\": \"{\\\"max_reads_per_bam\\\": \\\"250\\\", \\\"advanced_options_selector\\\": \\\"advanced\\\", \\\"extended_BAQ_computation\\\": \\\"False\\\", \\\"region_string\\\": \\\"\\\", \\\"output_per_sample_strand_bias_p_value\\\": \\\"False\\\", \\\"minimum_base_quality\\\": \\\"30\\\", \\\"disable_probabilistic_realignment\\\": \\\"False\\\", \\\"skip_anomalous_read_pairs\\\": \\\"False\\\", \\\"minimum_mapping_quality\\\": \\\"0\\\", \\\"output_per_sample_read_depth\\\": \\\"False\\\", \\\"__current_case__\\\": 0, \\\"position_list\\\": null, \\\"coefficient_for_downgrading\\\": \\\"0\\\"}\", \"reference_source\": \"{\\\"ref_file\\\": null, \\\"reference_source_selector\\\": \\\"history\\\", \\\"input_bams\\\": [{\\\"__index__\\\": 0, \\\"input_bam\\\": null}], \\\"__current_case__\\\": 1}\", \"chromInfo\": \"\\\"/home/galaxy/galaxy/galaxy-dist/tool-data/shared/ucsc/chrom/?.len\\\"\"}", + "tool_version": "0.0.3", + "type": "tool", + "user_outputs": [] + }, + "18": { + "annotation": "", + "id": 18, + "input_connections": { + "input_file": { + "id": 17, + "output_name": "output_mpileup" + } + }, + "inputs": [], + "name": "PileupVariant", + "outputs": [ + { + "name": "output_file", + "type": "pileup" + } + ], + "position": { + "left": 2458, + "top": 200 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "PileupVariant", + "tool_state": "{\"__page__\": 0, \"__rerun_remap_job_id__\": null, \"input_file\": \"null\"}", + "tool_version": "0.01", + "type": "tool", + "user_outputs": [] + }, + "19": { + "annotation": "", + "id": 19, + "input_connections": { + "input_listA": { + "id": 18, + "output_name": "output_file" + }, + "input_listB": { + "id": 3, + "output_name": "output" + } + }, + "inputs": [], + "name": "listfiltering", + "outputs": [ + { + "name": "output_file", + "type": "pileup" + } + ], + "position": { + "left": 2786, + "top": 200 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "toolshed.g2.bx.psu.edu/repos/mcharles/rapsodyn/listfiltering/0.01", + "tool_state": "{\"__page__\": 0, \"__rerun_remap_job_id__\": null, \"input_listB\": \"null\", \"nb_col\": \"\\\"2\\\"\", \"input_listA\": \"null\"}", + "tool_version": "0.01", + "type": "tool", + "user_outputs": [] + }, + "20": { + "annotation": "", + "id": 20, + "input_connections": { + "input_file": { + "id": 19, + "output_name": "output_file" + } + }, + "inputs": [], + "name": "SplitFile2x", + "outputs": [ + { + "name": "output_file1", + "type": "txt" + }, + { + "name": "output_file2", + "type": "txt" + } + ], + "position": { + "left": 3114, + "top": 200 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "SplitFile2x", + "tool_state": "{\"out_format\": \"\\\"pileup\\\"\", \"line_number\": \"\\\"4\\\"\", \"__page__\": 0, \"__rerun_remap_job_id__\": null, \"input_file\": \"null\"}", + "tool_version": "0.01", + "type": "tool", + "user_outputs": [] + }, + "21": { + "annotation": "", + "id": 21, + "input_connections": { + "input_assembly_file": { + "id": 0, + "output_name": "output" + }, + "input_variant_file": { + "id": 20, + "output_name": "output_file1" + } + }, + "inputs": [], + "name": "extractseq", + "outputs": [ + { + "name": "output_file", + "type": "fasta" + } + ], + "position": { + "left": 3442, + "top": 200 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "toolshed.g2.bx.psu.edu/repos/mcharles/rapsodyn/extractseq/0.03", + "tool_state": "{\"__page__\": 0, \"input_variant_file\": \"null\", \"__rerun_remap_job_id__\": null, \"window_length\": \"\\\"50\\\"\", \"input_assembly_file\": \"null\"}", + "tool_version": "0.03", + "type": "tool", + "user_outputs": [] + }, + "22": { + "annotation": "", + "id": 22, + "input_connections": { + "input_assembly_file": { + "id": 0, + "output_name": "output" + }, + "input_variant_file": { + "id": 20, + "output_name": "output_file2" + } + }, + "inputs": [], + "name": "extractseq", + "outputs": [ + { + "name": "output_file", + "type": "fasta" + } + ], + "position": { + "left": 3442, + "top": 375 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "toolshed.g2.bx.psu.edu/repos/mcharles/rapsodyn/extractseq/0.03", + "tool_state": "{\"__page__\": 0, \"input_variant_file\": \"null\", \"__rerun_remap_job_id__\": null, \"window_length\": \"\\\"50\\\"\", \"input_assembly_file\": \"null\"}", + "tool_version": "0.03", + "type": "tool", + "user_outputs": [] + }, + "23": { + "annotation": "", + "id": 23, + "input_connections": { + "db_opts|histdb": { + "id": 7, + "output_name": "outfile" + }, + "query": { + "id": 21, + "output_name": "output_file" + } + }, + "inputs": [], + "name": "NCBI BLAST+ blastn", + "outputs": [ + { + "name": "output1", + "type": "tabular" + } + ], + "position": { + "left": 3770, + "top": 200 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "toolshed.g2.bx.psu.edu/repos/devteam/ncbi_blast_plus/ncbi_blastn_wrapper/0.1.00", + "tool_state": "{\"evalue_cutoff\": \"\\\"0.001\\\"\", \"__page__\": 0, \"adv_opts\": \"{\\\"identity_cutoff\\\": \\\"0.0\\\", \\\"adv_opts_selector\\\": \\\"advanced\\\", \\\"ungapped\\\": \\\"False\\\", \\\"filter_query\\\": \\\"False\\\", \\\"word_size\\\": \\\"0\\\", \\\"__current_case__\\\": 1, \\\"parse_deflines\\\": \\\"False\\\", \\\"strand\\\": \\\"-strand both\\\", \\\"max_hits\\\": \\\"0\\\"}\", \"__rerun_remap_job_id__\": null, \"blast_type\": \"\\\"megablast\\\"\", \"db_opts\": \"{\\\"db_opts_selector\\\": \\\"histdb\\\", \\\"subject\\\": \\\"\\\", \\\"histdb\\\": null, \\\"__current_case__\\\": 1, \\\"database\\\": \\\"\\\"}\", \"output\": \"{\\\"out_format\\\": \\\"5\\\", \\\"__current_case__\\\": 3}\", \"query\": \"null\"}", + "tool_version": "0.1.00", + "type": "tool", + "user_outputs": [] + }, + "24": { + "annotation": "", + "id": 24, + "input_connections": { + "db_opts|histdb": { + "id": 7, + "output_name": "outfile" + }, + "query": { + "id": 22, + "output_name": "output_file" + } + }, + "inputs": [], + "name": "NCBI BLAST+ blastn", + "outputs": [ + { + "name": "output1", + "type": "tabular" + } + ], + "position": { + "left": 3770, + "top": 341 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "toolshed.g2.bx.psu.edu/repos/devteam/ncbi_blast_plus/ncbi_blastn_wrapper/0.1.00", + "tool_state": "{\"evalue_cutoff\": \"\\\"0.001\\\"\", \"__page__\": 0, \"adv_opts\": \"{\\\"identity_cutoff\\\": \\\"0.0\\\", \\\"adv_opts_selector\\\": \\\"advanced\\\", \\\"ungapped\\\": \\\"False\\\", \\\"filter_query\\\": \\\"False\\\", \\\"word_size\\\": \\\"0\\\", \\\"__current_case__\\\": 1, \\\"parse_deflines\\\": \\\"False\\\", \\\"strand\\\": \\\"-strand both\\\", \\\"max_hits\\\": \\\"0\\\"}\", \"__rerun_remap_job_id__\": null, \"blast_type\": \"\\\"megablast\\\"\", \"db_opts\": \"{\\\"db_opts_selector\\\": \\\"histdb\\\", \\\"subject\\\": \\\"\\\", \\\"histdb\\\": null, \\\"__current_case__\\\": 1, \\\"database\\\": \\\"\\\"}\", \"output\": \"{\\\"out_format\\\": \\\"5\\\", \\\"__current_case__\\\": 3}\", \"query\": \"null\"}", + "tool_version": "0.1.00", + "type": "tool", + "user_outputs": [] + }, + "25": { + "annotation": "", + "id": 25, + "input_connections": { + "input_blastxml_file": { + "id": 23, + "output_name": "output1" + }, + "input_variant_file": { + "id": 20, + "output_name": "output_file1" + } + }, + "inputs": [], + "name": "mpileupfilteronblastxml", + "outputs": [ + { + "name": "output_file", + "type": "pileup" + } + ], + "position": { + "left": 4072, + "top": 200 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "toolshed.g2.bx.psu.edu/repos/mcharles/rapsodyn/mpileupfilteronblastxml/0.03", + "tool_state": "{\"input_variant_file\": \"null\", \"__page__\": 0, \"__rerun_remap_job_id__\": null, \"input_blastxml_file\": \"null\", \"window_length\": \"\\\"50\\\"\", \"nb_mismatch_max\": \"\\\"3\\\"\"}", + "tool_version": "0.03", + "type": "tool", + "user_outputs": [] + }, + "26": { + "annotation": "", + "id": 26, + "input_connections": { + "input_blastxml_file": { + "id": 24, + "output_name": "output1" + }, + "input_variant_file": { + "id": 20, + "output_name": "output_file2" + } + }, + "inputs": [], + "name": "mpileupfilteronblastxml", + "outputs": [ + { + "name": "output_file", + "type": "pileup" + } + ], + "position": { + "left": 4072, + "top": 375 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "toolshed.g2.bx.psu.edu/repos/mcharles/rapsodyn/mpileupfilteronblastxml/0.03", + "tool_state": "{\"input_variant_file\": \"null\", \"__page__\": 0, \"__rerun_remap_job_id__\": null, \"input_blastxml_file\": \"null\", \"window_length\": \"\\\"50\\\"\", \"nb_mismatch_max\": \"\\\"3\\\"\"}", + "tool_version": "0.03", + "type": "tool", + "user_outputs": [] + }, + "27": { + "annotation": "", + "id": 27, + "input_connections": { + "input_file1": { + "id": 25, + "output_name": "output_file" + }, + "input_file2": { + "id": 26, + "output_name": "output_file" + } + }, + "inputs": [], + "name": "MergeFile2x", + "outputs": [ + { + "name": "output_file", + "type": "txt" + } + ], + "position": { + "left": 4400, + "top": 200 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "MergeFile2x", + "tool_state": "{\"out_format\": \"\\\"pileup\\\"\", \"__page__\": 0, \"__rerun_remap_job_id__\": null, \"input_file1\": \"null\", \"input_file2\": \"null\"}", + "tool_version": "0.01", + "type": "tool", + "user_outputs": [] + }, + "28": { + "annotation": "", + "id": 28, + "input_connections": { + "input_file": { + "id": 27, + "output_name": "output_file" + } + }, + "inputs": [], + "name": "mpileupfilterandstat", + "outputs": [ + { + "name": "output_file", + "type": "pileup" + }, + { + "name": "log_file", + "type": "txt" + } + ], + "position": { + "left": 4728, + "top": 200 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "mpileupfilterandstat", + "tool_state": "{\"__page__\": 0, \"input_file\": \"null\", \"min_frequency\": \"\\\"0.8\\\"\", \"do_stat\": \"\\\"YES\\\"\", \"min_depth\": \"\\\"2\\\"\", \"__rerun_remap_job_id__\": null, \"min_forward_and_reverse\": \"\\\"0\\\"\", \"max_depth\": \"\\\"100\\\"\", \"min_distance\": \"\\\"0\\\"\"}", + "tool_version": "0.07", + "type": "tool", + "user_outputs": [] + } + } +} \ No newline at end of file diff -r edddaa8ab855 -r e8e6b962c1f2 rapsodyn/Galaxy-Workflow-rapsosnp_v1.13_-_4x.ga --- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/rapsodyn/Galaxy-Workflow-rapsosnp_v1.13_-_4x.ga Fri Sep 05 06:12:10 2014 -0400 @@ -0,0 +1,1439 @@ +{ + "a_galaxy_workflow": "true", + "annotation": "", + "format-version": "0.1", + "name": "rapsosnp v1.13 - 4x", + "steps": { + "0": { + "annotation": "", + "id": 0, + "input_connections": {}, + "inputs": [ + { + "description": "", + "name": "READ1" + } + ], + "name": "Input dataset", + "outputs": [], + "position": { + "left": 214.76666259765625, + "top": 518.5 + }, + "tool_errors": null, + "tool_id": null, + "tool_state": "{\"name\": \"READ1\"}", + "tool_version": null, + "type": "data_input", + "user_outputs": [] + }, + "1": { + "annotation": "", + "id": 1, + "input_connections": {}, + "inputs": [ + { + "description": "", + "name": "READ2" + } + ], + "name": "Input dataset", + "outputs": [], + "position": { + "left": 219.76666259765625, + "top": 900.5 + }, + "tool_errors": null, + "tool_id": null, + "tool_state": "{\"name\": \"READ2\"}", + "tool_version": null, + "type": "data_input", + "user_outputs": [] + }, + "2": { + "annotation": "", + "id": 2, + "input_connections": {}, + "inputs": [ + { + "description": "", + "name": "ASSEMBLY" + } + ], + "name": "Input dataset", + "outputs": [], + "position": { + "left": 2639.7666015625, + "top": 382.5 + }, + "tool_errors": null, + "tool_id": null, + "tool_state": "{\"name\": \"ASSEMBLY\"}", + "tool_version": null, + "type": "data_input", + "user_outputs": [] + }, + "3": { + "annotation": "", + "id": 3, + "input_connections": {}, + "inputs": [ + { + "description": "", + "name": "DUBIUS" + } + ], + "name": "Input dataset", + "outputs": [], + "position": { + "left": 2449.7666015625, + "top": 1200.5 + }, + "tool_errors": null, + "tool_id": null, + "tool_state": "{\"name\": \"DUBIUS\"}", + "tool_version": null, + "type": "data_input", + "user_outputs": [] + }, + "4": { + "annotation": "", + "id": 4, + "input_connections": { + "input_file": { + "id": 0, + "output_name": "output" + } + }, + "inputs": [], + "name": "SplitFile4x", + "outputs": [ + { + "name": "output_file1", + "type": "txt" + }, + { + "name": "output_file2", + "type": "txt" + }, + { + "name": "output_file3", + "type": "txt" + }, + { + "name": "output_file4", + "type": "txt" + } + ], + "position": { + "left": 516.7666625976562, + "top": 398 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "SplitFile4x", + "tool_state": "{\"out_format\": \"\\\"fastq\\\"\", \"line_number\": \"\\\"4\\\"\", \"__page__\": 0, \"__rerun_remap_job_id__\": null, \"input_file\": \"null\"}", + "tool_version": "0.01", + "type": "tool", + "user_outputs": [] + }, + "5": { + "annotation": "", + "id": 5, + "input_connections": { + "input_file": { + "id": 1, + "output_name": "output" + } + }, + "inputs": [], + "name": "SplitFile4x", + "outputs": [ + { + "name": "output_file1", + "type": "txt" + }, + { + "name": "output_file2", + "type": "txt" + }, + { + "name": "output_file3", + "type": "txt" + }, + { + "name": "output_file4", + "type": "txt" + } + ], + "position": { + "left": 519.7666625976562, + "top": 716 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "SplitFile4x", + "tool_state": "{\"out_format\": \"\\\"fastq\\\"\", \"line_number\": \"\\\"4\\\"\", \"__page__\": 0, \"__rerun_remap_job_id__\": null, \"input_file\": \"null\"}", + "tool_version": "0.01", + "type": "tool", + "user_outputs": [] + }, + "6": { + "annotation": "", + "id": 6, + "input_connections": { + "input_fasta": { + "id": 2, + "output_name": "output" + } + }, + "inputs": [], + "name": "fastaGroomerForMakeBlastdb", + "outputs": [ + { + "name": "output_fasta", + "type": "fasta" + } + ], + "position": { + "left": 2796.7666015625, + "top": 200 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "toolshed.g2.bx.psu.edu/repos/mcharles/genephys/fastaGroomerForMakeBlastdb/0.01", + "tool_state": "{\"input_fasta\": \"null\", \"__rerun_remap_job_id__\": null, \"__page__\": 0}", + "tool_version": "0.01", + "type": "tool", + "user_outputs": [] + }, + "7": { + "annotation": "", + "id": 7, + "input_connections": { + "input_read1_file": { + "id": 4, + "output_name": "output_file1" + }, + "input_read2_file": { + "id": 5, + "output_name": "output_file1" + } + }, + "inputs": [], + "name": "PrepareFastqLight", + "outputs": [ + { + "name": "output_read1_file", + "type": "fastqsanger" + }, + { + "name": "output_read2_file", + "type": "fastqsanger" + } + ], + "position": { + "left": 891.7666625976562, + "top": 277 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "PrepareFastqLight", + "tool_state": "{\"__page__\": 0, \"quality_type\": \"\\\"illumina\\\"\", \"min_length\": \"\\\"30\\\"\", \"__rerun_remap_job_id__\": null, \"min_quality\": \"\\\"30\\\"\", \"input_read1_file\": \"null\", \"input_read2_file\": \"null\"}", + "tool_version": "0.01", + "type": "tool", + "user_outputs": [] + }, + "8": { + "annotation": "", + "id": 8, + "input_connections": { + "input_read1_file": { + "id": 4, + "output_name": "output_file2" + }, + "input_read2_file": { + "id": 5, + "output_name": "output_file2" + } + }, + "inputs": [], + "name": "PrepareFastqLight", + "outputs": [ + { + "name": "output_read1_file", + "type": "fastqsanger" + }, + { + "name": "output_read2_file", + "type": "fastqsanger" + } + ], + "position": { + "left": 900.7666625976562, + "top": 518 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "PrepareFastqLight", + "tool_state": "{\"__page__\": 0, \"quality_type\": \"\\\"illumina\\\"\", \"min_length\": \"\\\"30\\\"\", \"__rerun_remap_job_id__\": null, \"min_quality\": \"\\\"30\\\"\", \"input_read1_file\": \"null\", \"input_read2_file\": \"null\"}", + "tool_version": "0.01", + "type": "tool", + "user_outputs": [] + }, + "9": { + "annotation": "", + "id": 9, + "input_connections": { + "input_read1_file": { + "id": 4, + "output_name": "output_file3" + }, + "input_read2_file": { + "id": 5, + "output_name": "output_file3" + } + }, + "inputs": [], + "name": "PrepareFastqLight", + "outputs": [ + { + "name": "output_read1_file", + "type": "fastqsanger" + }, + { + "name": "output_read2_file", + "type": "fastqsanger" + } + ], + "position": { + "left": 890.7666625976562, + "top": 743 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "PrepareFastqLight", + "tool_state": "{\"__page__\": 0, \"quality_type\": \"\\\"illumina\\\"\", \"min_length\": \"\\\"30\\\"\", \"__rerun_remap_job_id__\": null, \"min_quality\": \"\\\"30\\\"\", \"input_read1_file\": \"null\", \"input_read2_file\": \"null\"}", + "tool_version": "0.01", + "type": "tool", + "user_outputs": [] + }, + "10": { + "annotation": "", + "id": 10, + "input_connections": { + "input_read1_file": { + "id": 4, + "output_name": "output_file4" + }, + "input_read2_file": { + "id": 5, + "output_name": "output_file4" + } + }, + "inputs": [], + "name": "PrepareFastqLight", + "outputs": [ + { + "name": "output_read1_file", + "type": "fastqsanger" + }, + { + "name": "output_read2_file", + "type": "fastqsanger" + } + ], + "position": { + "left": 894.7666625976562, + "top": 971 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "PrepareFastqLight", + "tool_state": "{\"__page__\": 0, \"quality_type\": \"\\\"illumina\\\"\", \"min_length\": \"\\\"30\\\"\", \"__rerun_remap_job_id__\": null, \"min_quality\": \"\\\"30\\\"\", \"input_read1_file\": \"null\", \"input_read2_file\": \"null\"}", + "tool_version": "0.01", + "type": "tool", + "user_outputs": [] + }, + "11": { + "annotation": "", + "id": 11, + "input_connections": { + "input_file": { + "id": 6, + "output_name": "output_fasta" + } + }, + "inputs": [], + "name": "NCBI BLAST+ makeblastdb", + "outputs": [ + { + "name": "outfile", + "type": "data" + } + ], + "position": { + "left": 3514.2666015625, + "top": 223 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "toolshed.g2.bx.psu.edu/repos/devteam/ncbi_blast_plus/ncbi_makeblastdb/0.1.00", + "tool_state": "{\"__page__\": 0, \"mask_data_file\": \"null\", \"input_file\": \"null\", \"dbtype\": \"\\\"nucl\\\"\", \"__rerun_remap_job_id__\": null, \"hash_index\": \"\\\"True\\\"\", \"tax\": \"{\\\"taxselect\\\": \\\"\\\", \\\"__current_case__\\\": 0}\", \"title\": \"\\\"\\\"\", \"parse_seqids\": \"\\\"False\\\"\"}", + "tool_version": "0.1.00", + "type": "tool", + "user_outputs": [] + }, + "12": { + "annotation": "", + "id": 12, + "input_connections": { + "genomeSource|ownFile": { + "id": 2, + "output_name": "output" + }, + "paired|input1": { + "id": 7, + "output_name": "output_read1_file" + }, + "paired|input2": { + "id": 7, + "output_name": "output_read2_file" + } + }, + "inputs": [], + "name": "Map with BWA for Illumina", + "outputs": [ + { + "name": "output", + "type": "sam" + } + ], + "position": { + "left": 1355.7666015625, + "top": 358.5 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "toolshed.g2.bx.psu.edu/repos/devteam/bwa_wrappers/bwa_wrapper/1.2.3", + "tool_state": "{\"genomeSource\": \"{\\\"refGenomeSource\\\": \\\"history\\\", \\\"ownFile\\\": null, \\\"__current_case__\\\": 1}\", \"__page__\": 0, \"__rerun_remap_job_id__\": null, \"paired\": \"{\\\"input2\\\": null, \\\"sPaired\\\": \\\"paired\\\", \\\"input1\\\": null, \\\"__current_case__\\\": 1}\", \"params\": \"{\\\"__current_case__\\\": 0, \\\"source_select\\\": \\\"pre_set\\\"}\", \"suppressHeader\": \"\\\"False\\\"\", \"chromInfo\": \"\\\"/home/galaxy/galaxy/galaxy-dist/tool-data/shared/ucsc/chrom/?.len\\\"\"}", + "tool_version": "1.2.3", + "type": "tool", + "user_outputs": [] + }, + "13": { + "annotation": "", + "id": 13, + "input_connections": { + "genomeSource|ownFile": { + "id": 2, + "output_name": "output" + }, + "paired|input1": { + "id": 8, + "output_name": "output_read1_file" + }, + "paired|input2": { + "id": 8, + "output_name": "output_read2_file" + } + }, + "inputs": [], + "name": "Map with BWA for Illumina", + "outputs": [ + { + "name": "output", + "type": "sam" + } + ], + "position": { + "left": 1349.7666015625, + "top": 540.5 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "toolshed.g2.bx.psu.edu/repos/devteam/bwa_wrappers/bwa_wrapper/1.2.3", + "tool_state": "{\"genomeSource\": \"{\\\"refGenomeSource\\\": \\\"history\\\", \\\"ownFile\\\": null, \\\"__current_case__\\\": 1}\", \"__page__\": 0, \"__rerun_remap_job_id__\": null, \"paired\": \"{\\\"input2\\\": null, \\\"sPaired\\\": \\\"paired\\\", \\\"input1\\\": null, \\\"__current_case__\\\": 1}\", \"params\": \"{\\\"__current_case__\\\": 0, \\\"source_select\\\": \\\"pre_set\\\"}\", \"suppressHeader\": \"\\\"False\\\"\", \"chromInfo\": \"\\\"/home/galaxy/galaxy/galaxy-dist/tool-data/shared/ucsc/chrom/?.len\\\"\"}", + "tool_version": "1.2.3", + "type": "tool", + "user_outputs": [] + }, + "14": { + "annotation": "", + "id": 14, + "input_connections": { + "genomeSource|ownFile": { + "id": 2, + "output_name": "output" + }, + "paired|input1": { + "id": 9, + "output_name": "output_read1_file" + }, + "paired|input2": { + "id": 9, + "output_name": "output_read2_file" + } + }, + "inputs": [], + "name": "Map with BWA for Illumina", + "outputs": [ + { + "name": "output", + "type": "sam" + } + ], + "position": { + "left": 1349.7666015625, + "top": 709.5 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "toolshed.g2.bx.psu.edu/repos/devteam/bwa_wrappers/bwa_wrapper/1.2.3", + "tool_state": "{\"genomeSource\": \"{\\\"refGenomeSource\\\": \\\"history\\\", \\\"ownFile\\\": null, \\\"__current_case__\\\": 1}\", \"__page__\": 0, \"__rerun_remap_job_id__\": null, \"paired\": \"{\\\"input2\\\": null, \\\"sPaired\\\": \\\"paired\\\", \\\"input1\\\": null, \\\"__current_case__\\\": 1}\", \"params\": \"{\\\"__current_case__\\\": 0, \\\"source_select\\\": \\\"pre_set\\\"}\", \"suppressHeader\": \"\\\"False\\\"\", \"chromInfo\": \"\\\"/home/galaxy/galaxy/galaxy-dist/tool-data/shared/ucsc/chrom/?.len\\\"\"}", + "tool_version": "1.2.3", + "type": "tool", + "user_outputs": [] + }, + "15": { + "annotation": "", + "id": 15, + "input_connections": { + "genomeSource|ownFile": { + "id": 2, + "output_name": "output" + }, + "paired|input1": { + "id": 10, + "output_name": "output_read1_file" + }, + "paired|input2": { + "id": 10, + "output_name": "output_read2_file" + } + }, + "inputs": [], + "name": "Map with BWA for Illumina", + "outputs": [ + { + "name": "output", + "type": "sam" + } + ], + "position": { + "left": 1349.7666015625, + "top": 889.5 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "toolshed.g2.bx.psu.edu/repos/devteam/bwa_wrappers/bwa_wrapper/1.2.3", + "tool_state": "{\"genomeSource\": \"{\\\"refGenomeSource\\\": \\\"history\\\", \\\"ownFile\\\": null, \\\"__current_case__\\\": 1}\", \"__page__\": 0, \"__rerun_remap_job_id__\": null, \"paired\": \"{\\\"input2\\\": null, \\\"sPaired\\\": \\\"paired\\\", \\\"input1\\\": null, \\\"__current_case__\\\": 1}\", \"params\": \"{\\\"__current_case__\\\": 0, \\\"source_select\\\": \\\"pre_set\\\"}\", \"suppressHeader\": \"\\\"False\\\"\", \"chromInfo\": \"\\\"/home/galaxy/galaxy/galaxy-dist/tool-data/shared/ucsc/chrom/?.len\\\"\"}", + "tool_version": "1.2.3", + "type": "tool", + "user_outputs": [] + }, + "16": { + "annotation": "", + "id": 16, + "input_connections": { + "input_sam_file": { + "id": 12, + "output_name": "output" + } + }, + "inputs": [], + "name": "filtersam_mapped_and_unique", + "outputs": [ + { + "name": "output_file", + "type": "sam" + } + ], + "position": { + "left": 1628.7666015625, + "top": 506.5 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "toolshed.g2.bx.psu.edu/repos/mcharles/rapsodyn/filtersam_mapped_and_unique/0.01", + "tool_state": "{\"__page__\": 0, \"__rerun_remap_job_id__\": null, \"chromInfo\": \"\\\"/home/galaxy/galaxy/galaxy-dist/tool-data/shared/ucsc/chrom/?.len\\\"\", \"input_sam_file\": \"null\"}", + "tool_version": "0.01", + "type": "tool", + "user_outputs": [] + }, + "17": { + "annotation": "", + "id": 17, + "input_connections": { + "input_sam_file": { + "id": 13, + "output_name": "output" + } + }, + "inputs": [], + "name": "filtersam_mapped_and_unique", + "outputs": [ + { + "name": "output_file", + "type": "sam" + } + ], + "position": { + "left": 1628.7666015625, + "top": 651.5 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "toolshed.g2.bx.psu.edu/repos/mcharles/rapsodyn/filtersam_mapped_and_unique/0.01", + "tool_state": "{\"__page__\": 0, \"__rerun_remap_job_id__\": null, \"chromInfo\": \"\\\"/home/galaxy/galaxy/galaxy-dist/tool-data/shared/ucsc/chrom/?.len\\\"\", \"input_sam_file\": \"null\"}", + "tool_version": "0.01", + "type": "tool", + "user_outputs": [] + }, + "18": { + "annotation": "", + "id": 18, + "input_connections": { + "input_sam_file": { + "id": 14, + "output_name": "output" + } + }, + "inputs": [], + "name": "filtersam_mapped_and_unique", + "outputs": [ + { + "name": "output_file", + "type": "sam" + } + ], + "position": { + "left": 1633.7666015625, + "top": 790.5 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "toolshed.g2.bx.psu.edu/repos/mcharles/rapsodyn/filtersam_mapped_and_unique/0.01", + "tool_state": "{\"__page__\": 0, \"__rerun_remap_job_id__\": null, \"chromInfo\": \"\\\"/home/galaxy/galaxy/galaxy-dist/tool-data/shared/ucsc/chrom/?.len\\\"\", \"input_sam_file\": \"null\"}", + "tool_version": "0.01", + "type": "tool", + "user_outputs": [] + }, + "19": { + "annotation": "", + "id": 19, + "input_connections": { + "input_sam_file": { + "id": 15, + "output_name": "output" + } + }, + "inputs": [], + "name": "filtersam_mapped_and_unique", + "outputs": [ + { + "name": "output_file", + "type": "sam" + } + ], + "position": { + "left": 1632.7666015625, + "top": 944.5 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "toolshed.g2.bx.psu.edu/repos/mcharles/rapsodyn/filtersam_mapped_and_unique/0.01", + "tool_state": "{\"__page__\": 0, \"__rerun_remap_job_id__\": null, \"chromInfo\": \"\\\"/home/galaxy/galaxy/galaxy-dist/tool-data/shared/ucsc/chrom/?.len\\\"\", \"input_sam_file\": \"null\"}", + "tool_version": "0.01", + "type": "tool", + "user_outputs": [] + }, + "20": { + "annotation": "", + "id": 20, + "input_connections": { + "source|input1": { + "id": 16, + "output_name": "output_file" + }, + "source|ref_file": { + "id": 2, + "output_name": "output" + } + }, + "inputs": [], + "name": "SAM-to-BAM", + "outputs": [ + { + "name": "output1", + "type": "bam" + } + ], + "position": { + "left": 1952.7666015625, + "top": 551.5 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "toolshed.g2.bx.psu.edu/repos/devteam/sam_to_bam/sam_to_bam/1.1.4", + "tool_state": "{\"source\": \"{\\\"index_source\\\": \\\"history\\\", \\\"ref_file\\\": null, \\\"input1\\\": null, \\\"__current_case__\\\": 1}\", \"__rerun_remap_job_id__\": null, \"chromInfo\": \"\\\"/home/galaxy/galaxy/galaxy-dist/tool-data/shared/ucsc/chrom/?.len\\\"\", \"__page__\": 0}", + "tool_version": "1.1.4", + "type": "tool", + "user_outputs": [] + }, + "21": { + "annotation": "", + "id": 21, + "input_connections": { + "source|input1": { + "id": 17, + "output_name": "output_file" + }, + "source|ref_file": { + "id": 2, + "output_name": "output" + } + }, + "inputs": [], + "name": "SAM-to-BAM", + "outputs": [ + { + "name": "output1", + "type": "bam" + } + ], + "position": { + "left": 1948.7666015625, + "top": 698.5 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "toolshed.g2.bx.psu.edu/repos/devteam/sam_to_bam/sam_to_bam/1.1.4", + "tool_state": "{\"source\": \"{\\\"index_source\\\": \\\"history\\\", \\\"ref_file\\\": null, \\\"input1\\\": null, \\\"__current_case__\\\": 1}\", \"__rerun_remap_job_id__\": null, \"chromInfo\": \"\\\"/home/galaxy/galaxy/galaxy-dist/tool-data/shared/ucsc/chrom/?.len\\\"\", \"__page__\": 0}", + "tool_version": "1.1.4", + "type": "tool", + "user_outputs": [] + }, + "22": { + "annotation": "", + "id": 22, + "input_connections": { + "source|input1": { + "id": 18, + "output_name": "output_file" + }, + "source|ref_file": { + "id": 2, + "output_name": "output" + } + }, + "inputs": [], + "name": "SAM-to-BAM", + "outputs": [ + { + "name": "output1", + "type": "bam" + } + ], + "position": { + "left": 1948.7666015625, + "top": 836.5 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "toolshed.g2.bx.psu.edu/repos/devteam/sam_to_bam/sam_to_bam/1.1.4", + "tool_state": "{\"source\": \"{\\\"index_source\\\": \\\"history\\\", \\\"ref_file\\\": null, \\\"input1\\\": null, \\\"__current_case__\\\": 1}\", \"__rerun_remap_job_id__\": null, \"chromInfo\": \"\\\"/home/galaxy/galaxy/galaxy-dist/tool-data/shared/ucsc/chrom/?.len\\\"\", \"__page__\": 0}", + "tool_version": "1.1.4", + "type": "tool", + "user_outputs": [] + }, + "23": { + "annotation": "", + "id": 23, + "input_connections": { + "source|input1": { + "id": 19, + "output_name": "output_file" + }, + "source|ref_file": { + "id": 2, + "output_name": "output" + } + }, + "inputs": [], + "name": "SAM-to-BAM", + "outputs": [ + { + "name": "output1", + "type": "bam" + } + ], + "position": { + "left": 1948.7666015625, + "top": 988.5 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "toolshed.g2.bx.psu.edu/repos/devteam/sam_to_bam/sam_to_bam/1.1.4", + "tool_state": "{\"source\": \"{\\\"index_source\\\": \\\"history\\\", \\\"ref_file\\\": null, \\\"input1\\\": null, \\\"__current_case__\\\": 1}\", \"__rerun_remap_job_id__\": null, \"chromInfo\": \"\\\"/home/galaxy/galaxy/galaxy-dist/tool-data/shared/ucsc/chrom/?.len\\\"\", \"__page__\": 0}", + "tool_version": "1.1.4", + "type": "tool", + "user_outputs": [] + }, + "24": { + "annotation": "", + "id": 24, + "input_connections": { + "input1": { + "id": 20, + "output_name": "output1" + }, + "input2": { + "id": 21, + "output_name": "output1" + }, + "inputs_0|input": { + "id": 22, + "output_name": "output1" + }, + "inputs_1|input": { + "id": 23, + "output_name": "output1" + } + }, + "inputs": [], + "name": "Merge BAM Files", + "outputs": [ + { + "name": "output1", + "type": "bam" + }, + { + "name": "outlog", + "type": "txt" + } + ], + "position": { + "left": 2265.7666015625, + "top": 833.5 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "toolshed.g2.bx.psu.edu/repos/devteam/sam_merge/sam_merge2/1.1.2", + "tool_state": "{\"input2\": \"null\", \"inputs\": \"[{\\\"__index__\\\": 0, \\\"input\\\": null}, {\\\"__index__\\\": 1, \\\"input\\\": null}]\", \"input1\": \"null\", \"title\": \"\\\"merge bam\\\"\", \"__page__\": 0, \"__rerun_remap_job_id__\": null, \"mergeSD\": \"\\\"True\\\"\", \"chromInfo\": \"\\\"/home/galaxy/galaxy/galaxy-dist/tool-data/shared/ucsc/chrom/?.len\\\"\"}", + "tool_version": "1.1.2", + "type": "tool", + "user_outputs": [] + }, + "25": { + "annotation": "", + "id": 25, + "input_connections": { + "reference_source|input_bams_0|input_bam": { + "id": 24, + "output_name": "output1" + }, + "reference_source|ref_file": { + "id": 2, + "output_name": "output" + } + }, + "inputs": [], + "name": "MPileup", + "outputs": [ + { + "name": "output_mpileup", + "type": "pileup" + }, + { + "name": "output_log", + "type": "txt" + } + ], + "position": { + "left": 2401.7666015625, + "top": 559.5 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "toolshed.g2.bx.psu.edu/repos/devteam/samtools_mpileup/samtools_mpileup/0.0.3", + "tool_state": "{\"__page__\": 0, \"genotype_likelihood_computation_type\": \"{\\\"genotype_likelihood_computation_type_selector\\\": \\\"do_not_perform_genotype_likelihood_computation\\\", \\\"__current_case__\\\": 1}\", \"__rerun_remap_job_id__\": null, \"advanced_options\": \"{\\\"max_reads_per_bam\\\": \\\"250\\\", \\\"advanced_options_selector\\\": \\\"advanced\\\", \\\"extended_BAQ_computation\\\": \\\"False\\\", \\\"region_string\\\": \\\"\\\", \\\"output_per_sample_strand_bias_p_value\\\": \\\"False\\\", \\\"minimum_base_quality\\\": \\\"30\\\", \\\"disable_probabilistic_realignment\\\": \\\"False\\\", \\\"skip_anomalous_read_pairs\\\": \\\"False\\\", \\\"minimum_mapping_quality\\\": \\\"0\\\", \\\"output_per_sample_read_depth\\\": \\\"False\\\", \\\"__current_case__\\\": 0, \\\"position_list\\\": null, \\\"coefficient_for_downgrading\\\": \\\"0\\\"}\", \"reference_source\": \"{\\\"ref_file\\\": null, \\\"reference_source_selector\\\": \\\"history\\\", \\\"input_bams\\\": [{\\\"__index__\\\": 0, \\\"input_bam\\\": null}], \\\"__current_case__\\\": 1}\", \"chromInfo\": \"\\\"/home/galaxy/galaxy/galaxy-dist/tool-data/shared/ucsc/chrom/?.len\\\"\"}", + "tool_version": "0.0.3", + "type": "tool", + "user_outputs": [] + }, + "26": { + "annotation": "", + "id": 26, + "input_connections": { + "input_file": { + "id": 25, + "output_name": "output_mpileup" + } + }, + "inputs": [], + "name": "PileupVariant", + "outputs": [ + { + "name": "output_file", + "type": "pileup" + } + ], + "position": { + "left": 2694.7666015625, + "top": 1060 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "PileupVariant", + "tool_state": "{\"__page__\": 0, \"__rerun_remap_job_id__\": null, \"input_file\": \"null\"}", + "tool_version": "0.01", + "type": "tool", + "user_outputs": [] + }, + "27": { + "annotation": "", + "id": 27, + "input_connections": { + "input_listA": { + "id": 26, + "output_name": "output_file" + }, + "input_listB": { + "id": 3, + "output_name": "output" + } + }, + "inputs": [], + "name": "listfiltering", + "outputs": [ + { + "name": "output_file", + "type": "pileup" + } + ], + "position": { + "left": 2778.7666015625, + "top": 752 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "toolshed.g2.bx.psu.edu/repos/mcharles/rapsodyn/listfiltering/0.01", + "tool_state": "{\"__page__\": 0, \"__rerun_remap_job_id__\": null, \"input_listB\": \"null\", \"nb_col\": \"\\\"2\\\"\", \"input_listA\": \"null\"}", + "tool_version": "0.01", + "type": "tool", + "user_outputs": [] + }, + "28": { + "annotation": "", + "id": 28, + "input_connections": { + "input_file": { + "id": 27, + "output_name": "output_file" + } + }, + "inputs": [], + "name": "SplitFile4x", + "outputs": [ + { + "name": "output_file1", + "type": "txt" + }, + { + "name": "output_file2", + "type": "txt" + }, + { + "name": "output_file3", + "type": "txt" + }, + { + "name": "output_file4", + "type": "txt" + } + ], + "position": { + "left": 3154, + "top": 702.38330078125 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "SplitFile4x", + "tool_state": "{\"out_format\": \"\\\"pileup\\\"\", \"line_number\": \"\\\"4\\\"\", \"__page__\": 0, \"__rerun_remap_job_id__\": null, \"input_file\": \"null\"}", + "tool_version": "0.01", + "type": "tool", + "user_outputs": [] + }, + "29": { + "annotation": "", + "id": 29, + "input_connections": { + "input_assembly_file": { + "id": 2, + "output_name": "output" + }, + "input_variant_file": { + "id": 28, + "output_name": "output_file1" + } + }, + "inputs": [], + "name": "extractseq", + "outputs": [ + { + "name": "output_file", + "type": "fasta" + } + ], + "position": { + "left": 3512, + "top": 507.3833312988281 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "toolshed.g2.bx.psu.edu/repos/mcharles/rapsodyn/extractseq/0.03", + "tool_state": "{\"__page__\": 0, \"input_variant_file\": \"null\", \"__rerun_remap_job_id__\": null, \"window_length\": \"\\\"50\\\"\", \"input_assembly_file\": \"null\"}", + "tool_version": "0.03", + "type": "tool", + "user_outputs": [] + }, + "30": { + "annotation": "", + "id": 30, + "input_connections": { + "input_assembly_file": { + "id": 2, + "output_name": "output" + }, + "input_variant_file": { + "id": 28, + "output_name": "output_file2" + } + }, + "inputs": [], + "name": "extractseq", + "outputs": [ + { + "name": "output_file", + "type": "fasta" + } + ], + "position": { + "left": 3517, + "top": 676.38330078125 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "toolshed.g2.bx.psu.edu/repos/mcharles/rapsodyn/extractseq/0.03", + "tool_state": "{\"__page__\": 0, \"input_variant_file\": \"null\", \"__rerun_remap_job_id__\": null, \"window_length\": \"\\\"50\\\"\", \"input_assembly_file\": \"null\"}", + "tool_version": "0.03", + "type": "tool", + "user_outputs": [] + }, + "31": { + "annotation": "", + "id": 31, + "input_connections": { + "input_assembly_file": { + "id": 2, + "output_name": "output" + }, + "input_variant_file": { + "id": 28, + "output_name": "output_file3" + } + }, + "inputs": [], + "name": "extractseq", + "outputs": [ + { + "name": "output_file", + "type": "fasta" + } + ], + "position": { + "left": 3518, + "top": 845.38330078125 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "toolshed.g2.bx.psu.edu/repos/mcharles/rapsodyn/extractseq/0.03", + "tool_state": "{\"__page__\": 0, \"input_variant_file\": \"null\", \"__rerun_remap_job_id__\": null, \"window_length\": \"\\\"50\\\"\", \"input_assembly_file\": \"null\"}", + "tool_version": "0.03", + "type": "tool", + "user_outputs": [] + }, + "32": { + "annotation": "", + "id": 32, + "input_connections": { + "input_assembly_file": { + "id": 2, + "output_name": "output" + }, + "input_variant_file": { + "id": 28, + "output_name": "output_file4" + } + }, + "inputs": [], + "name": "extractseq", + "outputs": [ + { + "name": "output_file", + "type": "fasta" + } + ], + "position": { + "left": 3525, + "top": 1012.38330078125 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "toolshed.g2.bx.psu.edu/repos/mcharles/rapsodyn/extractseq/0.03", + "tool_state": "{\"__page__\": 0, \"input_variant_file\": \"null\", \"__rerun_remap_job_id__\": null, \"window_length\": \"\\\"50\\\"\", \"input_assembly_file\": \"null\"}", + "tool_version": "0.03", + "type": "tool", + "user_outputs": [] + }, + "33": { + "annotation": "", + "id": 33, + "input_connections": { + "db_opts|histdb": { + "id": 11, + "output_name": "outfile" + }, + "query": { + "id": 29, + "output_name": "output_file" + } + }, + "inputs": [], + "name": "NCBI BLAST+ blastn", + "outputs": [ + { + "name": "output1", + "type": "tabular" + } + ], + "position": { + "left": 3959.7666015625, + "top": 491 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "toolshed.g2.bx.psu.edu/repos/devteam/ncbi_blast_plus/ncbi_blastn_wrapper/0.1.00", + "tool_state": "{\"evalue_cutoff\": \"\\\"0.001\\\"\", \"__page__\": 0, \"adv_opts\": \"{\\\"identity_cutoff\\\": \\\"0.0\\\", \\\"adv_opts_selector\\\": \\\"advanced\\\", \\\"ungapped\\\": \\\"False\\\", \\\"filter_query\\\": \\\"False\\\", \\\"word_size\\\": \\\"0\\\", \\\"__current_case__\\\": 1, \\\"parse_deflines\\\": \\\"False\\\", \\\"strand\\\": \\\"-strand both\\\", \\\"max_hits\\\": \\\"0\\\"}\", \"__rerun_remap_job_id__\": null, \"blast_type\": \"\\\"megablast\\\"\", \"db_opts\": \"{\\\"db_opts_selector\\\": \\\"histdb\\\", \\\"subject\\\": \\\"\\\", \\\"histdb\\\": null, \\\"__current_case__\\\": 1, \\\"database\\\": \\\"\\\"}\", \"output\": \"{\\\"out_format\\\": \\\"5\\\", \\\"__current_case__\\\": 3}\", \"query\": \"null\"}", + "tool_version": "0.1.00", + "type": "tool", + "user_outputs": [] + }, + "34": { + "annotation": "", + "id": 34, + "input_connections": { + "db_opts|histdb": { + "id": 11, + "output_name": "outfile" + }, + "query": { + "id": 30, + "output_name": "output_file" + } + }, + "inputs": [], + "name": "NCBI BLAST+ blastn", + "outputs": [ + { + "name": "output1", + "type": "tabular" + } + ], + "position": { + "left": 3955.7666015625, + "top": 675 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "toolshed.g2.bx.psu.edu/repos/devteam/ncbi_blast_plus/ncbi_blastn_wrapper/0.1.00", + "tool_state": "{\"evalue_cutoff\": \"\\\"0.001\\\"\", \"__page__\": 0, \"adv_opts\": \"{\\\"identity_cutoff\\\": \\\"0.0\\\", \\\"adv_opts_selector\\\": \\\"advanced\\\", \\\"ungapped\\\": \\\"False\\\", \\\"filter_query\\\": \\\"False\\\", \\\"word_size\\\": \\\"0\\\", \\\"__current_case__\\\": 1, \\\"parse_deflines\\\": \\\"False\\\", \\\"strand\\\": \\\"-strand both\\\", \\\"max_hits\\\": \\\"0\\\"}\", \"__rerun_remap_job_id__\": null, \"blast_type\": \"\\\"megablast\\\"\", \"db_opts\": \"{\\\"db_opts_selector\\\": \\\"histdb\\\", \\\"subject\\\": \\\"\\\", \\\"histdb\\\": null, \\\"__current_case__\\\": 1, \\\"database\\\": \\\"\\\"}\", \"output\": \"{\\\"out_format\\\": \\\"5\\\", \\\"__current_case__\\\": 3}\", \"query\": \"null\"}", + "tool_version": "0.1.00", + "type": "tool", + "user_outputs": [] + }, + "35": { + "annotation": "", + "id": 35, + "input_connections": { + "db_opts|histdb": { + "id": 11, + "output_name": "outfile" + }, + "query": { + "id": 31, + "output_name": "output_file" + } + }, + "inputs": [], + "name": "NCBI BLAST+ blastn", + "outputs": [ + { + "name": "output1", + "type": "tabular" + } + ], + "position": { + "left": 3965.7666015625, + "top": 848 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "toolshed.g2.bx.psu.edu/repos/devteam/ncbi_blast_plus/ncbi_blastn_wrapper/0.1.00", + "tool_state": "{\"evalue_cutoff\": \"\\\"0.001\\\"\", \"__page__\": 0, \"adv_opts\": \"{\\\"identity_cutoff\\\": \\\"0.0\\\", \\\"adv_opts_selector\\\": \\\"advanced\\\", \\\"ungapped\\\": \\\"False\\\", \\\"filter_query\\\": \\\"False\\\", \\\"word_size\\\": \\\"0\\\", \\\"__current_case__\\\": 1, \\\"parse_deflines\\\": \\\"False\\\", \\\"strand\\\": \\\"-strand both\\\", \\\"max_hits\\\": \\\"0\\\"}\", \"__rerun_remap_job_id__\": null, \"blast_type\": \"\\\"megablast\\\"\", \"db_opts\": \"{\\\"db_opts_selector\\\": \\\"histdb\\\", \\\"subject\\\": \\\"\\\", \\\"histdb\\\": null, \\\"__current_case__\\\": 1, \\\"database\\\": \\\"\\\"}\", \"output\": \"{\\\"out_format\\\": \\\"5\\\", \\\"__current_case__\\\": 3}\", \"query\": \"null\"}", + "tool_version": "0.1.00", + "type": "tool", + "user_outputs": [] + }, + "36": { + "annotation": "", + "id": 36, + "input_connections": { + "db_opts|histdb": { + "id": 11, + "output_name": "outfile" + }, + "query": { + "id": 32, + "output_name": "output_file" + } + }, + "inputs": [], + "name": "NCBI BLAST+ blastn", + "outputs": [ + { + "name": "output1", + "type": "tabular" + } + ], + "position": { + "left": 3971.7666015625, + "top": 1016 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "toolshed.g2.bx.psu.edu/repos/devteam/ncbi_blast_plus/ncbi_blastn_wrapper/0.1.00", + "tool_state": "{\"evalue_cutoff\": \"\\\"0.001\\\"\", \"__page__\": 0, \"adv_opts\": \"{\\\"identity_cutoff\\\": \\\"0.0\\\", \\\"adv_opts_selector\\\": \\\"advanced\\\", \\\"ungapped\\\": \\\"False\\\", \\\"filter_query\\\": \\\"False\\\", \\\"word_size\\\": \\\"0\\\", \\\"__current_case__\\\": 1, \\\"parse_deflines\\\": \\\"False\\\", \\\"strand\\\": \\\"-strand both\\\", \\\"max_hits\\\": \\\"0\\\"}\", \"__rerun_remap_job_id__\": null, \"blast_type\": \"\\\"megablast\\\"\", \"db_opts\": \"{\\\"db_opts_selector\\\": \\\"histdb\\\", \\\"subject\\\": \\\"\\\", \\\"histdb\\\": null, \\\"__current_case__\\\": 1, \\\"database\\\": \\\"\\\"}\", \"output\": \"{\\\"out_format\\\": \\\"5\\\", \\\"__current_case__\\\": 3}\", \"query\": \"null\"}", + "tool_version": "0.1.00", + "type": "tool", + "user_outputs": [] + }, + "37": { + "annotation": "", + "id": 37, + "input_connections": { + "input_blastxml_file": { + "id": 33, + "output_name": "output1" + }, + "input_variant_file": { + "id": 28, + "output_name": "output_file1" + } + }, + "inputs": [], + "name": "mpileupfilteronblastxml", + "outputs": [ + { + "name": "output_file", + "type": "pileup" + } + ], + "position": { + "left": 4260.5, + "top": 478.3833312988281 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "toolshed.g2.bx.psu.edu/repos/mcharles/rapsodyn/mpileupfilteronblastxml/0.03", + "tool_state": "{\"input_variant_file\": \"null\", \"__page__\": 0, \"__rerun_remap_job_id__\": null, \"input_blastxml_file\": \"null\", \"window_length\": \"\\\"50\\\"\", \"nb_mismatch_max\": \"\\\"3\\\"\"}", + "tool_version": "0.03", + "type": "tool", + "user_outputs": [] + }, + "38": { + "annotation": "", + "id": 38, + "input_connections": { + "input_blastxml_file": { + "id": 34, + "output_name": "output1" + }, + "input_variant_file": { + "id": 28, + "output_name": "output_file2" + } + }, + "inputs": [], + "name": "mpileupfilteronblastxml", + "outputs": [ + { + "name": "output_file", + "type": "pileup" + } + ], + "position": { + "left": 4262.5, + "top": 656.38330078125 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "toolshed.g2.bx.psu.edu/repos/mcharles/rapsodyn/mpileupfilteronblastxml/0.03", + "tool_state": "{\"input_variant_file\": \"null\", \"__page__\": 0, \"__rerun_remap_job_id__\": null, \"input_blastxml_file\": \"null\", \"window_length\": \"\\\"50\\\"\", \"nb_mismatch_max\": \"\\\"3\\\"\"}", + "tool_version": "0.03", + "type": "tool", + "user_outputs": [] + }, + "39": { + "annotation": "", + "id": 39, + "input_connections": { + "input_blastxml_file": { + "id": 35, + "output_name": "output1" + }, + "input_variant_file": { + "id": 28, + "output_name": "output_file3" + } + }, + "inputs": [], + "name": "mpileupfilteronblastxml", + "outputs": [ + { + "name": "output_file", + "type": "pileup" + } + ], + "position": { + "left": 4264.5, + "top": 832.38330078125 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "toolshed.g2.bx.psu.edu/repos/mcharles/rapsodyn/mpileupfilteronblastxml/0.03", + "tool_state": "{\"input_variant_file\": \"null\", \"__page__\": 0, \"__rerun_remap_job_id__\": null, \"input_blastxml_file\": \"null\", \"window_length\": \"\\\"50\\\"\", \"nb_mismatch_max\": \"\\\"3\\\"\"}", + "tool_version": "0.03", + "type": "tool", + "user_outputs": [] + }, + "40": { + "annotation": "", + "id": 40, + "input_connections": { + "input_blastxml_file": { + "id": 36, + "output_name": "output1" + }, + "input_variant_file": { + "id": 28, + "output_name": "output_file4" + } + }, + "inputs": [], + "name": "mpileupfilteronblastxml", + "outputs": [ + { + "name": "output_file", + "type": "pileup" + } + ], + "position": { + "left": 4267.5, + "top": 1005.38330078125 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "toolshed.g2.bx.psu.edu/repos/mcharles/rapsodyn/mpileupfilteronblastxml/0.03", + "tool_state": "{\"input_variant_file\": \"null\", \"__page__\": 0, \"__rerun_remap_job_id__\": null, \"input_blastxml_file\": \"null\", \"window_length\": \"\\\"50\\\"\", \"nb_mismatch_max\": \"\\\"3\\\"\"}", + "tool_version": "0.03", + "type": "tool", + "user_outputs": [] + }, + "41": { + "annotation": "", + "id": 41, + "input_connections": { + "input_file1": { + "id": 37, + "output_name": "output_file" + }, + "input_file2": { + "id": 38, + "output_name": "output_file" + }, + "input_file3": { + "id": 39, + "output_name": "output_file" + }, + "input_file4": { + "id": 40, + "output_name": "output_file" + } + }, + "inputs": [], + "name": "MergeFile4x", + "outputs": [ + { + "name": "output_file", + "type": "txt" + } + ], + "position": { + "left": 4700, + "top": 637.38330078125 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "MergeFile4x", + "tool_state": "{\"out_format\": \"\\\"pileup\\\"\", \"__page__\": 0, \"__rerun_remap_job_id__\": null, \"input_file4\": \"null\", \"input_file1\": \"null\", \"input_file2\": \"null\", \"input_file3\": \"null\"}", + "tool_version": "0.01", + "type": "tool", + "user_outputs": [] + }, + "42": { + "annotation": "", + "id": 42, + "input_connections": { + "input_file": { + "id": 41, + "output_name": "output_file" + } + }, + "inputs": [], + "name": "mpileupfilterandstat", + "outputs": [ + { + "name": "output_file", + "type": "pileup" + }, + { + "name": "log_file", + "type": "txt" + } + ], + "position": { + "left": 5056, + "top": 682.38330078125 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "mpileupfilterandstat", + "tool_state": "{\"__page__\": 0, \"input_file\": \"null\", \"min_frequency\": \"\\\"0.8\\\"\", \"do_stat\": \"\\\"YES\\\"\", \"min_depth\": \"\\\"2\\\"\", \"__rerun_remap_job_id__\": null, \"min_forward_and_reverse\": \"\\\"0\\\"\", \"max_depth\": \"\\\"100\\\"\", \"min_distance\": \"\\\"0\\\"\"}", + "tool_version": "0.07", + "type": "tool", + "user_outputs": [] + } + } +} \ No newline at end of file diff -r edddaa8ab855 -r e8e6b962c1f2 rapsodyn/PileupVariant.pl --- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/rapsodyn/PileupVariant.pl Fri Sep 05 06:12:10 2014 -0400 @@ -0,0 +1,19 @@ +#!/usr/bin/perl +use strict; + +my $inputfile = $ARGV[0]; +open(IF, $inputfile) or die("Can't open $inputfile\n"); + +#Extraction des variants +my $nb_line=0; +while (my $line=){ + my $test = $line; + $test =~ s/\$//g; #the read start at this position + $test =~ s/\^.//g; #the read end at this position followed by quality char + my @field = split(/\s+/,$test); + + if ($field[4]=~/[ATGCN]/i){ + print $line; + } +} +close(IF); diff -r edddaa8ab855 -r e8e6b962c1f2 rapsodyn/PileupVariant.xml --- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/rapsodyn/PileupVariant.xml Fri Sep 05 06:12:10 2014 -0400 @@ -0,0 +1,18 @@ + +Keep only mpileup line with variant + + PileupVariant.pl $input_file > $output_file + + + + + + + + + + + + + + diff -r edddaa8ab855 -r e8e6b962c1f2 rapsodyn/PrepareFastqLight.pl --- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/rapsodyn/PrepareFastqLight.pl Fri Sep 05 06:12:10 2014 -0400 @@ -0,0 +1,420 @@ +#!/usr/bin/perl +use strict; +use warnings; + +my $read1 = $ARGV[0]; +my $read2 = $ARGV[1]; + +my $output1 = $ARGV[2]; +my $output2 = $ARGV[3]; + +my $TYPE = $ARGV[4]; +my $MIN_LENGTH = $ARGV[5]; +my $MIN_QUALITY = $ARGV[6]; + +my $VERBOSE = $ARGV[7]; + +if (!$VERBOSE){ + $VERBOSE ="OFF"; +} + +open(READ1, $read1) or die ("Can't open $read1\n"); +open(READ2, $read2) or die ("Can't open $read2\n"); +open(OUT1, ">$output1") or die ("Can't open $output1\n"); +open(OUT2, ">$output2") or die ("Can't open $output2\n"); + + +my $error1=0; +my $error2=0; +my $error3=0; +my $error4=0; +my $error5=0; +my $error6=0; +my $error7=0; +my $error8=0; +my $error9=0; +my $error10=0; + +while (my $ligne1_r1 =){ + my $ligne2_r1 =; + my $ligne3_r1 =; + my $ligne4_r1 =; + my $ligne1_r2 =; + my $ligne2_r2 =; + my $ligne3_r2 =; + my $ligne4_r2 =; + +#@ 1 sec + if ((!$ligne1_r1)||(!$ligne2_r1)||(!$ligne3_r1)||(!$ligne4_r1)||(!$ligne1_r2)||(!$ligne2_r2)||(!$ligne3_r2)||(!$ligne4_r2)){ + if ($VERBOSE eq "ON"){ + print "Error in file format"; + if ($ligne1_r1){print $ligne1_r1;} + if ($ligne2_r1){print $ligne2_r1;} + if ($ligne3_r1){print $ligne3_r1;} + if ($ligne4_r1){print $ligne4_r1;} + if ($ligne1_r2){print $ligne1_r2;} + if ($ligne2_r2){print $ligne2_r2;} + if ($ligne3_r2){print $ligne3_r2;} + if ($ligne4_r2){print $ligne4_r2;} + print "\n"; + } + $error1++; + } + elsif(($ligne1_r1 !~/^\@/)||($ligne1_r2 !~/^\@/)||($ligne3_r1 !~/^\+/)||($ligne3_r2 !~/^\+/)){ + if ($VERBOSE eq "ON"){ + print "Error in header : format\n"; + print $ligne1_r1; + print $ligne2_r1; + print $ligne3_r1; + print $ligne4_r1; + print $ligne1_r2; + print $ligne2_r2; + print $ligne3_r2; + print $ligne4_r2; + print "\n"; + } + $error2++; + } +#@ 1 - 2 sec + else { + + my $length_seq1 = length($ligne2_r1); + my $length_qual1 =length($ligne4_r1); + my $seq1; + my $qual1; + + my $length_seq2 = length($ligne2_r2); + my $length_qual2 =length($ligne4_r2); + my $seq2; + my $qual2; + my $header1=""; + my $header2=""; + my $repheader1=""; + my $repheader2=""; + + if ($ligne1_r1 =~/^\@(.*?)\#/){ + $header1 = $1; + } + + if ($ligne3_r1 =~/^\+(.*?)\#/){ + $repheader1 = $1; + } + + if ($ligne1_r2 =~/^\@(.*?)\#/){ + $header2 = $1; + } + + if ($ligne3_r2 =~/^\+(.*?)\#/){ + $repheader2 = $1; + } +#@ 2 sec + + ### Verification de la coherence sequence /qualité @ 1 sec + if (($TYPE eq "illumina")&&((!$header1)||(!$header2)||(!$repheader1)||(!$repheader2))){ + if ($VERBOSE eq "ON"){ + print "Error in header : empty\n"; + print $ligne1_r1; + print $ligne2_r1; + print $ligne3_r1; + print $ligne4_r1; + print $ligne1_r2; + print $ligne2_r2; + print $ligne3_r2; + print $ligne4_r2; + print "\n"; + } + $error3++; + } + elsif (($TYPE eq "sanger")&&((!$header1)||(!$header2))){ + if ($VERBOSE eq "ON"){ + print "Error in header refgsd : empty\n"; + print $ligne1_r1; + print $ligne2_r1; + print $ligne3_r1; + print $ligne4_r1; + print $ligne1_r2; + print $ligne2_r2; + print $ligne3_r2; + print $ligne4_r2; + print "\n"; + } + $error3++; + } + elsif (($TYPE eq "illumina")&&(($header1 ne $repheader1)||($header2 ne $repheader2)||($header1 ne $header2))){ + if ($VERBOSE eq "ON"){ + print "Error in header : different\n"; + print $ligne1_r1; + print $ligne2_r1; + print $ligne3_r1; + print $ligne4_r1; + print $ligne1_r2; + print $ligne2_r2; + print $ligne3_r2; + print $ligne4_r2; + print "\n"; + } + $error4++; + } + elsif (($TYPE eq "sanger")&&($header1 ne $header2)){ + if ($VERBOSE eq "ON"){ + print "Error in header : different\n"; + print $ligne1_r1; + print $ligne2_r1; + print $ligne3_r1; + print $ligne4_r1; + print $ligne1_r2; + print $ligne2_r2; + print $ligne3_r2; + print $ligne4_r2; + print "\n"; + } + $error4++; + } + elsif (($length_seq1 != $length_qual1)||($length_seq2 != $length_qual2)){ + if ($VERBOSE eq "ON"){ + print "Error in seq/qual length\n"; + print $ligne1_r1; + print $ligne2_r1; + print $ligne3_r1; + print $ligne4_r1; + print $ligne1_r2; + print $ligne2_r2; + print $ligne3_r2; + print $ligne4_r2; + print "\n"; + } + $error5++; + } +#@ 1 - 2 sec + else { + ### Parsing sequence & qualité + if ($ligne2_r1 =~ /^([ATGCNX]+)\s*$/i){ + $seq1 = $1; + } + if ($ligne2_r2 =~ /^([ATGCNX]+)\s*$/i){ + $seq2 = $1; + } + if ($ligne4_r1 =~ /^(.*)\s*$/i){ + $qual1 = $1; + } + if ($ligne4_r2 =~ /^(.*)\s*$/i){ + $qual2 = $1; + } +#@ 2 sec + ### Verification du parsing et de la coherence sequence /qualité (n°2) + if ((!$seq1)||(!$seq2)||(!$qual1)||(!$qual2)){ + if ($VERBOSE eq "ON"){ + print "Error parsing seq / quality \n"; + print $ligne1_r1; + print $ligne2_r1; + print $ligne3_r1; + print $ligne4_r1; + print $ligne1_r2; + print $ligne2_r2; + print $ligne3_r2; + print $ligne4_r2; + print "\n"; + } + $error6++; + } + elsif ((length($seq1) != length($qual1))||(length($seq2) != length($qual2))){ + if ($VERBOSE eq "ON"){ + print "Error in seq/qual length after parsing\n"; + print $ligne1_r1; + print $ligne2_r1; + print $ligne3_r1; + print $ligne4_r1; + print $ligne1_r2; + print $ligne2_r2; + print $ligne3_r2; + print $ligne4_r2; + print "\n"; + } + $error7++; + } +#@ <1 sec + else { + my $fastq_lines_r1=""; + my $fastq_lines_r2=""; + $fastq_lines_r1 = &grooming_and_trimming($ligne1_r1,$seq1,$qual1); + if ($fastq_lines_r1){ + $fastq_lines_r2 = &grooming_and_trimming($ligne1_r2,$seq2,$qual2); + } + if ($fastq_lines_r2){ + print OUT1 $fastq_lines_r1; + print OUT2 $fastq_lines_r2; + } + } + } + + # print OUT1 $ligne1_r1; + # print OUT1 $ligne2_r1; + # print OUT1 $ligne3_r1; + # print OUT1 $ligne4_r1; + # print OUT2 $ligne1_r2; + # print OUT2 $ligne2_r2; + # print OUT2 $ligne3_r2; + # print OUT2 $ligne4_r2; + +#@ 7 sec + } +} + + + +close (READ1); +close (READ2); +close (OUT1); +close (OUT2); + + + + +sub grooming_and_trimming{ + my $header = shift; + my $seq = shift; + my $quality = shift; + my $quality_converted=""; + + my $startnoN = 0; + my $stopnoN = length($quality)-1; + + #print "SEQ :\n$seq\n"; + + my $chercheN = $seq; + my @bad_position; + my $current_index = index($chercheN,"N"); + my $abs_index = $current_index; + while ($current_index >=0){ + push (@bad_position,$abs_index); + + if ($current_index=0){ + my %coord=%{&extract_longer_string_coordinates_from_bad_position($startnoN,$stopnoN,\@bad_position)}; + $startnoN = $coord{"start"}; + $stopnoN = $coord{"stop"}; + } + my $lengthnoN = $stopnoN - $startnoN + 1; + my $seqnoN = substr($seq,$startnoN,$lengthnoN); + #print "$seqnoN\n"; + + if ($lengthnoN >= $MIN_LENGTH){ + my $startTrim = $startnoN; + my $stopTrim = $stopnoN; + + my $quality_converted=""; + my @bad_position; + + my @q = split(//,$quality); + #print "QUALITY\n"; + #print "$quality\n"; + for (my $i=0;$i<=$stopnoN;$i++){ + my $chr = $q[$i]; + my $num = ord($q[$i]); + if ($TYPE eq "illumina"){ + $num = $num -64+33; + $quality_converted .= chr($num); + } + + if ($num <$MIN_QUALITY + 64 - 33 ){ + push(@bad_position,$i+$startnoN); + } + } + if ($quality_converted){$quality = $quality_converted;} + #print "$quality\n"; + + + + if ($#bad_position>=0){ + # for (my $i=0;$i<=$#bad_position;$i++){ + # print $bad_position[$i]."\t"; + # } + # print "\n"; + my %coord=%{&extract_longer_string_coordinates_from_bad_position($startnoN,$stopnoN,\@bad_position)}; + $startTrim = $coord{"start"}; + $stopTrim = $coord{"stop"}; + #print "$startTrim .. $stopTrim\n"; + + } + my $lengthTrim = $stopTrim - $startTrim +1; + + + my $fastq_lines=""; + + if ($lengthTrim >= $MIN_LENGTH){ + $fastq_lines .= $header; + $fastq_lines .= substr($seq,$startTrim,$lengthTrim)."\n"; + $fastq_lines .= "+\n"; + $fastq_lines .= substr($quality,$startTrim,$lengthTrim)."\n"; + return $fastq_lines; + } + else { + return ""; + } + + + + } + else { + return ""; + } + + + # my @s = split(//,$seq); + # my $sanger_quality=""; + + + + + # return $sanger_quality; +} + +sub extract_longer_string_coordinates_from_bad_position{ + my $start=shift; + my $stop =shift; + my $refbad = shift; + my @bad_position = @$refbad; + my %coord; + + my $current_start = $start; + my $current_stop = $bad_position[0]-1; + if ($current_stop < $start){$current_stop = $start;} + + + #debut -> premier N + my $current_length = $current_stop - $current_start +1; + my $test_length; + + #entre les N + for (my $i=1;$i<=$#bad_position;$i++){ + $test_length = $bad_position[$i]+1-$bad_position[$i-1]-1; + if ( $test_length > $current_length){ + $current_start = $bad_position[$i-1]+1; + $current_stop = $bad_position[$i]-1; + $current_length = $current_stop - $current_start +1; + } + } + + #dernier N -> fin + $test_length = $stop-$bad_position[$#bad_position]+1; + if ( $test_length > $current_length){ + $current_start = $bad_position[$#bad_position]+1; + if ($current_start > $stop){$current_start=$stop;} + $current_stop = $stop; + } + $coord{"start"}=$current_start; + $coord{"stop"}= $current_stop; + $coord{"lenght"}=$current_stop-$current_start+1; + + return \%coord; +} diff -r edddaa8ab855 -r e8e6b962c1f2 rapsodyn/PrepareFastqLight.xml --- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/rapsodyn/PrepareFastqLight.xml Fri Sep 05 06:12:10 2014 -0400 @@ -0,0 +1,26 @@ + +Fastq preparation + + PrepareFastqLight.pl $input_read1_file $input_read2_file $output_read1_file $output_read2_file $quality_type $min_quality $min_length + + + + + + + + + + + + + + + + + + + + + + diff -r edddaa8ab855 -r e8e6b962c1f2 rapsodyn/SplitFile2x.pl --- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/rapsodyn/SplitFile2x.pl Fri Sep 05 06:12:10 2014 -0400 @@ -0,0 +1,54 @@ +#!/usr/bin/perl +use strict; +use warnings; + +my $in = $ARGV[0]; +my $NB_LINE_GROUP = $ARGV[1]; +my $NB_FILE = 2; +my $out1 = $ARGV[2]; +my $out2 = $ARGV[3]; + +if (!$NB_LINE_GROUP){ + $NB_LINE_GROUP=4; +} + +if (!$NB_FILE){ + $NB_FILE=2; +} + +my $file_nb = 1; + +my @in; +my $current_aggregate; +open(IN, $in) or die ("Can't open $in\n"); +while (my $line =){ + $current_aggregate = $line; + my $nb_line=1; + if ($NB_LINE_GROUP>1){ + while (my $add_line = ){ + $current_aggregate .= $add_line; + $nb_line++; + if ($nb_line == $NB_LINE_GROUP){ + last; + } + } + } + push (@in,$current_aggregate); +} +close (IN); + +open (OUT1,">$out1") or die ("Can't open $out1"); +open (OUT2,">$out2") or die ("Can't open $out2"); + +for (my $i=0;$i<=$#in;$i++){ + if ($i <= $#in/2){ + print OUT1 $in[$i]; + } + else { + print OUT2 $in[$i]; + } + + +} +close (OUT1); +close (OUT2); diff -r edddaa8ab855 -r e8e6b962c1f2 rapsodyn/SplitFile2x.xml --- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/rapsodyn/SplitFile2x.xml Fri Sep 05 06:12:10 2014 -0400 @@ -0,0 +1,44 @@ + +Split file in 2 + + SplitFile2x.pl $input_file $line_number $output_file1 $output_file2 + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + diff -r edddaa8ab855 -r e8e6b962c1f2 rapsodyn/SplitFile4x.pl --- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/rapsodyn/SplitFile4x.pl Fri Sep 05 06:12:10 2014 -0400 @@ -0,0 +1,70 @@ +#!/usr/bin/perl +use strict; +use warnings; + +my $in = $ARGV[0]; +my $NB_LINE_GROUP = $ARGV[1]; +my $NB_FILE = 4; +my $out1 = $ARGV[2]; +my $out2 = $ARGV[3]; +my $out3 = $ARGV[4]; +my $out4 = $ARGV[5]; + +if (!$NB_LINE_GROUP){ + $NB_LINE_GROUP=4; +} + +if (!$NB_FILE){ + $NB_FILE=4; +} + +my $file_nb = 1; + +my @in; +my $current_aggregate; +open(IN, $in) or die ("Can't open $in\n"); +while (my $line =){ + $current_aggregate = $line; + my $nb_line=1; + if ($NB_LINE_GROUP>1){ + while (my $add_line = ){ + $current_aggregate .= $add_line; + $nb_line++; + if ($nb_line == $NB_LINE_GROUP){ + last; + } + } + } + push (@in,$current_aggregate); +} +close (IN); + +open (OUT1,">$out1") or die ("Can't open $out1"); +open (OUT2,">$out2") or die ("Can't open $out2"); +open (OUT3,">$out3") or die ("Can't open $out3"); +open (OUT4,">$out4") or die ("Can't open $out4"); + + +for (my $i=0;$i<=$#in;$i++){ + if ($i <= $#in/4){ + print OUT1 $in[$i]; + } + elsif ($i <= $#in/2){ + print OUT2 $in[$i]; + } + elsif ($i <= $#in*3/4){ + print OUT3 $in[$i]; + } + elsif ($i <= $#in){ + print OUT4 $in[$i]; + } + else { + + } + + +} +close (OUT1); +close (OUT2); +close (OUT3); +close (OUT4); diff -r edddaa8ab855 -r e8e6b962c1f2 rapsodyn/SplitFile4x.xml --- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/rapsodyn/SplitFile4x.xml Fri Sep 05 06:12:10 2014 -0400 @@ -0,0 +1,62 @@ + +Split file in 4 + + SplitFile4x.pl $input_file $line_number $output_file1 $output_file2 $output_file3 $output_file4 + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + diff -r edddaa8ab855 -r e8e6b962c1f2 rapsodyn/mpileupfilter.pl --- a/rapsodyn/mpileupfilter.pl Thu Aug 21 08:39:15 2014 -0400 +++ /dev/null Thu Jan 01 00:00:00 1970 +0000 @@ -1,394 +0,0 @@ -#!/usr/bin/perl -use strict; -use Getopt::Long; - -# -# Filter a pileup file on forward/reverse presence and %read having the variant -# The error code -# 1 : multiple variant type detected insertion/deletion/mutation -# 1i : inconsistency in insertion -# 1d : inconsistency in deletion -# 1m : inconsistency in mutation -# 2 : insufficient depth -# 3 : insufficient variant frequency -# 4 : variant position not covered by forward and reverse reads -# 5 : variant with other variant in neighbourhood -# 6 : too much depth -# 8 : parsing error (couldn't parse the mpileup line correctly) -# 9 : parsing error (couldn't parse the readbase string correctly) - - -my $inputfile; -my $logfile; -my $MIN_DISTANCE=0; -my $MIN_VARIANTFREQUENCY=0; -my $MIN_FORWARDREVERSE=0; -my $MIN_DEPTH=0; -my $MAX_DEPTH=500; -my $VERBOSE=0; -my $ONLY_UNFILTERED_VARIANT="OFF"; - -if ($#ARGV<0){ - print "\n"; - print "perl 020_FilterPileupv6 -input_file [OPTION]\n"; - print "-input_file \tinputfile in mpileup format\n"; - print "-log_file \tlogfile containing discarded mpileup lines and the errorcode associated\n"; - print "-min_depth \tminimum depth required [1]\n"; - print "-max_depth \tmaximim depth (position with more coverage will be discarded) [100]\n"; - print "-min_frequency \tminimum variant frequency (0->1) [1] (default 1 => 100% reads show the variant at this position)\n"; - print "-min_distance \tminimum distance between variant [0]\n"; - print "-min_forward_and_reverse \tminimum number of reads in forward and reverse covering the variant required [0]\n"; - print "\n"; - exit(0); -} - -GetOptions ( -"input_file=s" => \$inputfile, -"log_file=s" => \$logfile, -"min_depth=i" => \$MIN_DEPTH, -"max_depth=i" => \$MAX_DEPTH, -"min_frequency=f" => \$MIN_VARIANTFREQUENCY, -"min_distance=i" => \$MIN_DISTANCE, -"min_forward_and_reverse=i" => \$MIN_FORWARDREVERSE, -"variant_only=s" => \$ONLY_UNFILTERED_VARIANT, -"v=i" => \$VERBOSE -) or die("Error in command line arguments\n"); - - -open(IF, $inputfile) or die("Can't open $inputfile\n"); - -my @tbl_line; -my @tbl_variant_position; -my @tbl_variant_chr; -my @tbl_variant_refbase; -my @tbl_variant_coverage; -my @tbl_variant_readbase_string; -my @tbl_variant_quality_string; - -#Extraction des variants -my $nb_line=0; -while (my $line=){ - $nb_line++; - if (($nb_line % 1000000 == 0)&&($VERBOSE==1)){ - print "$nb_line\n"; - } - my $error_code=0; - if ($line=~/(.*?)\s+(\d+)\s+([ATGCN])\s+(\d+)\s+(.*?)\s+(.*?)$/){ - my $current_chromosome = $1; - my $current_position = $2; - my $current_refbase = $3; - my $current_coverage = $4; - my $current_readbase_string = $5; - my $current_quality_string = $6; - - #Suppression of mPileUp special character - $current_readbase_string =~ s/\$//g; #the read start at this position - $current_readbase_string =~ s/\^.//g; #the read end at this position followed by quality char - - if ($current_readbase_string =~ /[ATGCNatgcn\d]/){ - push(@tbl_line,$line); - push(@tbl_variant_chr,$current_chromosome); - push(@tbl_variant_position,$current_position); - push(@tbl_variant_refbase,$current_refbase); - push(@tbl_variant_coverage,$current_coverage); - push(@tbl_variant_readbase_string,$current_readbase_string); - push(@tbl_variant_quality_string,$current_quality_string); - if ($ONLY_UNFILTERED_VARIANT eq "ON"){ - print $line; - } - - } - else { - #Position with no variant - } - - } - else { - #Error Parsing - print STDERR "$line #8"; - } -} -close(IF); - -if ($ONLY_UNFILTERED_VARIANT eq "ON"){ - exit(0); -} - -####Checking the distance between variant and other filter - -if ($logfile){ - open(LF,">$logfile") or die ("Cant't open $logfile\n"); -} - -for (my $i=0;$i<=$#tbl_line;$i++){ - # print "ligne : $tbl_line[$i]\n"; - - my $error_code=0; - if ($i==0){ - #Comparing $i and $i+1 for neighbourhood filter; - if ($#tbl_line>0){ - if (($tbl_variant_chr[$i+1] eq $tbl_variant_chr[$i])&&($tbl_variant_position[$i]+$MIN_DISTANCE>=$tbl_variant_position[$i+1])){ - $error_code=5; - chomp($tbl_line[$i]); - if ($logfile){ - print LF "$tbl_line[$i]\tcode:$error_code\n"; - } - next; - } - } - - #Additionnal filters - $error_code = check_error($tbl_variant_chr[$i],$tbl_variant_position[$i],$tbl_variant_refbase[$i],$tbl_variant_coverage[$i],$tbl_variant_readbase_string[$i]); - - } - else { - #Compairing $i and $i-1 for neighbourhood filter - if (($tbl_variant_chr[$i-1] eq $tbl_variant_chr[$i])&&($tbl_variant_position[$i-1]+$MIN_DISTANCE>=$tbl_variant_position[$i])){ - $error_code=5; - chomp($tbl_line[$i]); - if ($logfile){ - print LF "$tbl_line[$i]\tcode:$error_code\n"; - } - next; - } - else { - #Additionnal filters - $error_code = check_error($tbl_variant_chr[$i],$tbl_variant_position[$i],$tbl_variant_refbase[$i],$tbl_variant_coverage[$i],$tbl_variant_readbase_string[$i]); - } - } - if ($error_code == 0){ - print $tbl_line[$i]; - } - else { - chomp($tbl_line[$i]); - if ($logfile){ - print LF "$tbl_line[$i]\tcode:$error_code\n"; - } - } -} - -if ($logfile){ - close (LF); -} - -sub check_error{ - my $current_chromosome = shift; - my $current_position = shift; - my $current_refbase = shift; - my $current_coverage = shift; - my $current_readbase_string = shift; - - # print "test : $current_readbase_string\n"; - - - - #Extraction of insertions - - ################################################################## - # my @IN = $current_readbase_string =~ m/\+[0-9]+[ACGTNacgtn]+/g; - # my @DEL = $current_readbase_string =~ m/\-[0-9]+[ACGTNacgtn]+/g; - # print "IN : @IN\n"; - # print "DEL :@DEL\n"; - #$current_readbase_string=~s/[\+\-][0-9]+[ACGTNacgtn]+//g; - ################################################################## - #!!! marche pas : exemple .+1Ct. correspond a . / +1C / t /. mais le match de l'expression vire +1Ct - ################################################################## - - # => parcours de boucle - my @readbase = split(//,$current_readbase_string); - my $cleaned_readbase_string=""; - my @IN; - my @DEL; - my $current_IN=""; - my $current_DEL=""; - my $current_size=0; - - for (my $i=0;$i<=$#readbase;$i++){ - if ($readbase[$i] eq "+"){ - #Ouverture de IN - $current_IN="+"; - - #Recuperation de la taille - my $sub = substr $current_readbase_string,$i; - if ($sub=~/^\+(\d+)/){ - $current_size = $1; - } - my $remaining_size = $current_size; - while (($remaining_size>0)&&($i<=$#readbase)){ - $i++; - $current_IN.=$readbase[$i]; - if ($readbase[$i]=~ /[ATGCNatgcn]/){ - $remaining_size--; - } - } - push(@IN,$current_IN); - } - elsif ($readbase[$i] eq "-"){ - #Ouverture de DEL - $current_DEL="-"; - - #Recuperation de la taille - my $sub = substr $current_readbase_string,$i; - if ($sub=~/^\-(\d+)/){ - $current_size = $1; - } - my $remaining_size = $current_size; - while (($remaining_size>0)&&($i<=$#readbase)){ - $i++; - $current_DEL.=$readbase[$i]; - if ($readbase[$i]=~ /[ATGCNatgcn]/){ - $remaining_size--; - } - } - push(@DEL,$current_DEL); - - } - else { - #Ajout a la string - $cleaned_readbase_string .= $readbase[$i]; - } - } - - - # print "IN : @IN\n"; - # print "DEL :@DEL\n"; - # print "$cleaned_readbase_string\n"; - - my @current_readbase_array = split(//,$cleaned_readbase_string); - - #Filtering : error detection - - if ($#current_readbase_array+1 != $current_coverage){ - return 9; - #parsing error (couldn't parse the readbase string correctly) - } - elsif ($current_coverage<$MIN_DEPTH){ - return 2; - # 2 : insufficient depth - } - elsif ($current_coverage>$MAX_DEPTH){ - return 6; - # 6 : too much depth - } - else { - if ($#IN>=0){ - if (($cleaned_readbase_string=~/[ACGTNacgtn]/)){ - return 1; - # 1 : variant type overload (multiple variant type detected insertion/deletion/mutation) - } - else { - ########## TEST de coherence des insertions ################ - # for (my $i=0;$i<=$#IN;$i++){ - # if (uc($IN[0]) ne uc($IN[$i])){ - # print uc($IN[0]),"\n"; - # print uc($IN[$i]),"\n"; - # return "1i"; - # } - # } - ########################################################### - - if($#IN+1 < $current_coverage*$MIN_VARIANTFREQUENCY){ - return 3; - # 3 : insufficient variant frequency - } - } - } - elsif ($#DEL>=0){ - if (($cleaned_readbase_string=~/[ACGTNacgtn]/)){ - return 1; - # 1 : variant type overload (multiple variant type detected insertion/deletion/mutation) - } - else { - ########## TEST de coherence des deletions ################ - # for (my $i=0;$i<=$#DEL;$i++){ - # if (uc($DEL[0]) ne uc($DEL[$i])){ - # print uc($DEL[0]),"\n"; - # print uc($DEL[$i]),"\n"; - # return "1d"; - # } - # } - ########################################################### - - if($#DEL+1 < $current_coverage*$MIN_VARIANTFREQUENCY){ - return 3; - # 3 : insufficient variant frequency - } - } - } - else { - my $nbA=0; - $nbA++ while ($current_readbase_string =~ m/A/g); - my $nbC=0; - $nbC++ while ($current_readbase_string =~ m/C/g); - my $nbT=0; - $nbT++ while ($current_readbase_string =~ m/T/g); - my $nbG=0; - $nbG++ while ($current_readbase_string =~ m/G/g); - my $nbN=0; - $nbN++ while ($current_readbase_string =~ m/N/g); - my $nba=0; - $nba++ while ($current_readbase_string =~ m/a/g); - my $nbc=0; - $nbc++ while ($current_readbase_string =~ m/c/g); - my $nbt=0; - $nbt++ while ($current_readbase_string =~ m/t/g); - my $nbg=0; - $nbg++ while ($current_readbase_string =~ m/g/g); - my $nbn=0; - $nbn++ while ($current_readbase_string =~ m/n/g); - - if (($nbA+$nba>0)&&($nbT+$nbt+$nbG+$nbg+$nbC+$nbc+$nbN+$nbn>0)){ - return "1m"; - } - if (($nbT+$nbt>0)&&($nbA+$nba+$nbG+$nbg+$nbC+$nbc+$nbN+$nbn>0)){ - return "1m"; - } - if (($nbG+$nbg>0)&&($nbA+$nba+$nbT+$nbt+$nbC+$nbc+$nbN+$nbn>0)){ - return "1m"; - } - if (($nbC+$nbc>0)&&($nbA+$nba+$nbT+$nbt+$nbG+$nbg+$nbN+$nbn>0)){ - return "1m"; - } - if (($nbN+$nbn>0)&&($nbA+$nba+$nbT+$nbt+$nbG+$nbg+$nbC+$nbc>0)){ - return "1m"; - } - - if ($nbA+$nba >= $current_coverage*$MIN_VARIANTFREQUENCY){ - if (($nbA<$MIN_FORWARDREVERSE)||($nba<$MIN_FORWARDREVERSE)){ - return 4; - # 4 : variant position not covered by forward and reverse reads - } - } - elsif ($nbT+$nbt >= $current_coverage*$MIN_VARIANTFREQUENCY){ - if (($nbT<$MIN_FORWARDREVERSE)||($nbt<$MIN_FORWARDREVERSE)){ - return 4; - # 4 : variant position not covered by forward and reverse reads - } - } - elsif ($nbG+$nbg >= $current_coverage*$MIN_VARIANTFREQUENCY){ - if (($nbG<$MIN_FORWARDREVERSE)||($nbg<$MIN_FORWARDREVERSE)){ - return 4; - # 4 : variant position not covered by forward and reverse reads - } - } - elsif ($nbC+$nbc >= $current_coverage*$MIN_VARIANTFREQUENCY){ - if (($nbC<$MIN_FORWARDREVERSE)||($nbc<$MIN_FORWARDREVERSE)){ - return 4; - # 4 : variant position not covered by forward and reverse reads - } - } - elsif ($nbN+$nbn >= $current_coverage*$MIN_VARIANTFREQUENCY){ - if (($nbN<$MIN_FORWARDREVERSE)||($nbn<$MIN_FORWARDREVERSE)){ - return 4; - # 4 : variant position not covered by forward and reverse reads - } - } - else { - return 3; - # 3 : insufficient variant frequency - } - } - } - - return 0; -} diff -r edddaa8ab855 -r e8e6b962c1f2 rapsodyn/mpileupfilter.xml --- a/rapsodyn/mpileupfilter.xml Thu Aug 21 08:39:15 2014 -0400 +++ /dev/null Thu Jan 01 00:00:00 1970 +0000 @@ -1,23 +0,0 @@ - -Filter mpileup file entry - - mpileupfilter.pl -input_file $input_file -min_depth $min_depth -min_frequency $min_frequency -min_distance $min_distance -min_forward_and_reverse $min_forward_and_reverse -max_depth $max_depth > $output_file - - - - - - - - - - - - - - - - - - - diff -r edddaa8ab855 -r e8e6b962c1f2 rapsodyn/mpileupfilterandstat.pl --- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/rapsodyn/mpileupfilterandstat.pl Fri Sep 05 06:12:10 2014 -0400 @@ -0,0 +1,469 @@ +#!/usr/bin/perl +use strict; +use Getopt::Long; + +# +# Filter a pileup file on forward/reverse presence and %read having the variant +# The error code +# 1 : multiple variant type detected insertion/deletion/mutation +# 1i : inconsistency in insertion +# 1d : inconsistency in deletion +# 1m : inconsistency in mutation +# 2 : insufficient depth +# 3 : insufficient variant frequency +# 4 : variant position not covered by forward and reverse reads +# 5 : variant with other variant in neighbourhood +# 6 : too much depth +# 8 : parsing error (couldn't parse the mpileup line correctly) +# 9 : parsing error (couldn't parse the readbase string correctly) + + +my $inputfile; +my $logfile; +my $MIN_DISTANCE=0; +my $MIN_VARIANTFREQUENCY=0; +my $MIN_FORWARDREVERSE=0; +my $MIN_DEPTH=0; +my $MAX_DEPTH=500; +my $VERBOSE=0; +my $ONLY_UNFILTERED_VARIANT="OFF"; +my $DO_STAT="NO"; + +if ($#ARGV<0){ + print "\n"; + print "perl 020_FilterPileupv6 -input_file [OPTION]\n"; + print "-input_file \tinputfile in mpileup format\n"; + print "-log_file \tlogfile containing discarded mpileup lines and the errorcode associated\n"; + print "-min_depth \tminimum depth required [1]\n"; + print "-max_depth \tmaximim depth (position with more coverage will be discarded) [100]\n"; + print "-min_frequency \tminimum variant frequency (0->1) [1] (default 1 => 100% reads show the variant at this position)\n"; + print "-min_distance \tminimum distance between variant [0]\n"; + print "-min_forward_and_reverse \tminimum number of reads in forward and reverse covering the variant required [0]\n"; + print "\n"; + exit(0); +} + +GetOptions ( +"input_file=s" => \$inputfile, +"log_file=s" => \$logfile, +"min_depth=i" => \$MIN_DEPTH, +"max_depth=i" => \$MAX_DEPTH, +"min_frequency=f" => \$MIN_VARIANTFREQUENCY, +"min_distance=i" => \$MIN_DISTANCE, +"min_forward_and_reverse=i" => \$MIN_FORWARDREVERSE, +"variant_only=s" => \$ONLY_UNFILTERED_VARIANT, +"v=i" => \$VERBOSE, +"do_stat=s" => \$DO_STAT +) or die("Error in command line arguments\n"); + + +open(IF, $inputfile) or die("Can't open $inputfile\n"); + +my @tbl_line; +my %USR_PARAM; +$USR_PARAM{"min_depth"} = $MIN_DEPTH; +$USR_PARAM{"max_depth"} = $MAX_DEPTH; +$USR_PARAM{"min_freq"} = $MIN_VARIANTFREQUENCY; +$USR_PARAM{"min_dist"} = $MIN_DISTANCE; +$USR_PARAM{"min_fr"} = $MIN_FORWARDREVERSE; + + + +#Extraction des variants +my $nb_line=0; +while (my $line=){ + $nb_line++; + if (($nb_line % 1000000 == 0)&&($VERBOSE==1)){ + print "$nb_line\n"; + } + my $error_code=0; + if ($line=~/(.*?)\s+(\d+)\s+([ATGCN])\s+(\d+)\s+(.*?)\s+(.*?)$/){ + my $current_chromosome = $1; + my $current_position = $2; + my $current_refbase = $3; + my $current_coverage = $4; + my $current_readbase_string = $5; + my $current_quality_string = $6; + + #Suppression of mPileUp special character + $current_readbase_string =~ s/\$//g; #the read start at this position + $current_readbase_string =~ s/\^.//g; #the read end at this position followed by quality char + + if ($current_readbase_string =~ /[ATGCNatgcn\d]/){ + my %variant; + $variant{"line"} = $line; + $variant{"chr"} = $current_chromosome; + $variant{"pos"} = $current_position; + $variant{"refbase"} = $current_refbase; + $variant{"coverage"} = $current_coverage; + $variant{"readbase"} = $current_readbase_string; + $variant{"quality"} = $current_quality_string; + push(@tbl_line,\%variant); + + if ($ONLY_UNFILTERED_VARIANT eq "ON"){ + print $line; + } + + } + else { + #Position with no variant + } + + } + else { + #Error Parsing + print STDERR "$line #8"; + } +} +close(IF); + +if ($ONLY_UNFILTERED_VARIANT eq "ON"){ + exit(0); +} + +####Checking the distance between variant and other filter + + +my @error; +for (my $i=0;$i<=$#tbl_line;$i++){ + # print "ligne : $tbl_line[$i]\n"; + my $before=""; + my $after=""; + my %line = %{$tbl_line[$i]}; + + if ($tbl_line[$i-1]){ + $before = $tbl_line[$i-1]; + } + if ($tbl_line[$i+1]){ + $after = $tbl_line[$i+1]; + } + my $error_code = check_error($tbl_line[$i],$before,$after,\%USR_PARAM); + if ($error_code == 0){ + print $line{"line"}; + } + else { + push(@error,$error_code,"\t",$line{"line"}); + } +} + +### LOG +open(LF,">$logfile") or die ("Can't open $logfile\n"); + +if ($DO_STAT eq "YES"){ + for (my $idx_min_depth=2;$idx_min_depth<=32;$idx_min_depth = $idx_min_depth*2){ + for (my $idx_freq = 0.5;$idx_freq<=1;$idx_freq= $idx_freq+0.1){ + for (my $idx_dist=0;$idx_dist<=50;$idx_dist = $idx_dist + 50){ + for (my $idx_fr=0;$idx_fr<=1;$idx_fr++){ + my %stat_param; + $stat_param{"min_depth"}=$idx_min_depth; + $stat_param{"max_depth"}=250; + $stat_param{"min_freq"}=$idx_freq; + $stat_param{"min_fr"}=$idx_fr; + $stat_param{"min_dist"}=$idx_dist; + + print LF "#SNP = ",&test_check(\@tbl_line,\%stat_param),"\tdepth (min/max) = ",$stat_param{"min_depth"}," / ",$stat_param{"max_depth"},"\tmin_dist=",$stat_param{"min_dist"},"\tmin_freq=",$stat_param{"min_freq"},"\tmin_forwardreverse = ",$stat_param{"min_fr"},"\n"; + } + } + } + print "\n"; + } +} + + +for (my $i=0;$i<=$#error;$i++){ + print LF $error[$i]; +} +close (LF); + + + + +sub test_check{ + my $ref_tbl_line = shift; + my $ref_param = shift; + my @tbl_line = @$ref_tbl_line; + my %param = %$ref_param; + my $nb=0; + + for (my $i=0;$i<=$#tbl_line;$i++){ + my $before=""; + my $after=""; + my %line = %{$tbl_line[$i]}; + + if ($tbl_line[$i-1]){ + $before = $tbl_line[$i-1]; + } + if ($tbl_line[$i+1]){ + $after = $tbl_line[$i+1]; + } + my $error_code = check_error($tbl_line[$i],$before,$after,\%param); + if ($error_code == 0){ + $nb++; + } + } + + return $nb; +} + +sub check_error{ + my $refline = shift; + my %line = %$refline; + my $refbefore = shift; + my $refafter = shift; + my $refparam = shift; + my %param = %$refparam; + + + my $current_chromosome = $line{"chr"}; + my $current_position = $line{"pos"}; + my $current_refbase = $line{"refbase"}; + my $current_coverage = $line{"coverage"}; + my $current_readbase_string = $line{"readbase"}; + + + my $min_depth = $param{"min_depth"}; + my $max_depth = $param{"max_depth"}; + my $min_variant_frequency = $param{"min_freq"}; + my $min_forward_reverse = $param{"min_fr"}; + my $min_dist = $param{"min_dist"}; + + #Verification of neightbourhood + if ($refbefore){ + my %compareline = %$refbefore; + my $compare_chromosome = $compareline{"chr"}; + my $compare_position = $compareline{"pos"}; + my $compare_refbase = $compareline{"refbase"}; + my $compare_coverage = $compareline{"coverage"}; + my $compare_readbase_string = $compareline{"readbase"}; + + if (($current_chromosome eq $compare_chromosome )&&($compare_position + $min_dist >= $current_position)){ + return 5; + } + } + + if ($refafter){ + my %compareline = %$refafter; + my $compare_chromosome = $compareline{"chr"}; + my $compare_position = $compareline{"pos"}; + my $compare_refbase = $compareline{"refbase"}; + my $compare_coverage = $compareline{"coverage"}; + my $compare_readbase_string = $compareline{"readbase"}; + + if (($current_chromosome eq $compare_chromosome )&&($current_position + $min_dist >= $compare_position)){ + return 5; + } + } + + + + + #Extraction of insertions + + ################################################################## + # my @IN = $current_readbase_string =~ m/\+[0-9]+[ACGTNacgtn]+/g; + # my @DEL = $current_readbase_string =~ m/\-[0-9]+[ACGTNacgtn]+/g; + # print "IN : @IN\n"; + # print "DEL :@DEL\n"; + #$current_readbase_string=~s/[\+\-][0-9]+[ACGTNacgtn]+//g; + ################################################################## + #!!! marche pas : exemple .+1Ct. correspond a . / +1C / t /. mais le match de l'expression vire +1Ct + ################################################################## + + # => parcours de boucle + my @readbase = split(//,$current_readbase_string); + my $cleaned_readbase_string=""; + my @IN; + my @DEL; + my $current_IN=""; + my $current_DEL=""; + my $current_size=0; + + for (my $i=0;$i<=$#readbase;$i++){ + if ($readbase[$i] eq "+"){ + #Ouverture de IN + $current_IN="+"; + + #Recuperation de la taille + my $sub = substr $current_readbase_string,$i; + if ($sub=~/^\+(\d+)/){ + $current_size = $1; + } + my $remaining_size = $current_size; + while (($remaining_size>0)&&($i<=$#readbase)){ + $i++; + $current_IN.=$readbase[$i]; + if ($readbase[$i]=~ /[ATGCNatgcn]/){ + $remaining_size--; + } + } + push(@IN,$current_IN); + } + elsif ($readbase[$i] eq "-"){ + #Ouverture de DEL + $current_DEL="-"; + + #Recuperation de la taille + my $sub = substr $current_readbase_string,$i; + if ($sub=~/^\-(\d+)/){ + $current_size = $1; + } + my $remaining_size = $current_size; + while (($remaining_size>0)&&($i<=$#readbase)){ + $i++; + $current_DEL.=$readbase[$i]; + if ($readbase[$i]=~ /[ATGCNatgcn]/){ + $remaining_size--; + } + } + push(@DEL,$current_DEL); + + } + else { + #Ajout a la string + $cleaned_readbase_string .= $readbase[$i]; + } + } + + + # print "IN : @IN\n"; + # print "DEL :@DEL\n"; + # print "$cleaned_readbase_string\n"; + + my @current_readbase_array = split(//,$cleaned_readbase_string); + + #Filtering : error detection + + if ($#current_readbase_array+1 != $current_coverage){ + return 9; + #parsing error (couldn't parse the readbase string correctly) + } + elsif ($current_coverage<$min_depth){ + return 2; + # 2 : insufficient depth + } + elsif ($current_coverage>$max_depth){ + return 6; + # 6 : too much depth + } + else { + if ($#IN>=0){ + if (($cleaned_readbase_string=~/[ACGTNacgtn]/)){ + return 1; + # 1 : variant type overload (multiple variant type detected insertion/deletion/mutation) + } + else { + ########## TEST de coherence des insertions ################ + # for (my $i=0;$i<=$#IN;$i++){ + # if (uc($IN[0]) ne uc($IN[$i])){ + # print uc($IN[0]),"\n"; + # print uc($IN[$i]),"\n"; + # return "1i"; + # } + # } + ########################################################### + + if($#IN+1 < $current_coverage*$min_variant_frequency ){ + return 3; + # 3 : insufficient variant frequency + } + } + } + elsif ($#DEL>=0){ + if (($cleaned_readbase_string=~/[ACGTNacgtn]/)){ + return 1; + # 1 : variant type overload (multiple variant type detected insertion/deletion/mutation) + } + else { + ########## TEST de coherence des deletions ################ + # for (my $i=0;$i<=$#DEL;$i++){ + # if (uc($DEL[0]) ne uc($DEL[$i])){ + # print uc($DEL[0]),"\n"; + # print uc($DEL[$i]),"\n"; + # return "1d"; + # } + # } + ########################################################### + + if($#DEL+1 < $current_coverage*$min_variant_frequency){ + return 3; + # 3 : insufficient variant frequency + } + } + } + else { + my $nbA=0; + $nbA++ while ($current_readbase_string =~ m/A/g); + my $nbC=0; + $nbC++ while ($current_readbase_string =~ m/C/g); + my $nbT=0; + $nbT++ while ($current_readbase_string =~ m/T/g); + my $nbG=0; + $nbG++ while ($current_readbase_string =~ m/G/g); + my $nbN=0; + $nbN++ while ($current_readbase_string =~ m/N/g); + my $nba=0; + $nba++ while ($current_readbase_string =~ m/a/g); + my $nbc=0; + $nbc++ while ($current_readbase_string =~ m/c/g); + my $nbt=0; + $nbt++ while ($current_readbase_string =~ m/t/g); + my $nbg=0; + $nbg++ while ($current_readbase_string =~ m/g/g); + my $nbn=0; + $nbn++ while ($current_readbase_string =~ m/n/g); + + if (($nbA+$nba>0)&&($nbT+$nbt+$nbG+$nbg+$nbC+$nbc+$nbN+$nbn>0)){ + return "1m"; + } + if (($nbT+$nbt>0)&&($nbA+$nba+$nbG+$nbg+$nbC+$nbc+$nbN+$nbn>0)){ + return "1m"; + } + if (($nbG+$nbg>0)&&($nbA+$nba+$nbT+$nbt+$nbC+$nbc+$nbN+$nbn>0)){ + return "1m"; + } + if (($nbC+$nbc>0)&&($nbA+$nba+$nbT+$nbt+$nbG+$nbg+$nbN+$nbn>0)){ + return "1m"; + } + if (($nbN+$nbn>0)&&($nbA+$nba+$nbT+$nbt+$nbG+$nbg+$nbC+$nbc>0)){ + return "1m"; + } + + if ($nbA+$nba >= $current_coverage*$min_variant_frequency){ + if (($nbA<$min_forward_reverse)||($nba<$min_forward_reverse)){ + return 4; + # 4 : variant position not covered by forward and reverse reads + } + } + elsif ($nbT+$nbt >= $current_coverage*$min_variant_frequency){ + if (($nbT<$min_forward_reverse)||($nbt<$min_forward_reverse)){ + return 4; + # 4 : variant position not covered by forward and reverse reads + } + } + elsif ($nbG+$nbg >= $current_coverage*$min_variant_frequency){ + if (($nbG<$min_forward_reverse)||($nbg<$min_forward_reverse)){ + return 4; + # 4 : variant position not covered by forward and reverse reads + } + } + elsif ($nbC+$nbc >= $current_coverage*$min_variant_frequency){ + if (($nbC<$min_forward_reverse)||($nbc<$min_forward_reverse)){ + return 4; + # 4 : variant position not covered by forward and reverse reads + } + } + elsif ($nbN+$nbn >= $current_coverage*$min_variant_frequency){ + if (($nbN<$min_forward_reverse)||($nbn<$min_forward_reverse)){ + return 4; + # 4 : variant position not covered by forward and reverse reads + } + } + else { + return 3; + # 3 : insufficient variant frequency + } + } + } + + return 0; +} diff -r edddaa8ab855 -r e8e6b962c1f2 rapsodyn/mpileupfilterandstat.xml --- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/rapsodyn/mpileupfilterandstat.xml Fri Sep 05 06:12:10 2014 -0400 @@ -0,0 +1,29 @@ + +Filter mpileup file entry + + mpileupfilterandstat.pl -input_file $input_file -log_file $log_file -min_depth $min_depth -min_frequency $min_frequency -min_distance $min_distance -min_forward_and_reverse $min_forward_and_reverse -max_depth $max_depth -do_stat $do_stat> $output_file + + + + + + + + + + + + + + + + + + + + + + + + + diff -r edddaa8ab855 -r e8e6b962c1f2 rapsodyn/rapsodyn_stats.pl --- a/rapsodyn/rapsodyn_stats.pl Thu Aug 21 08:39:15 2014 -0400 +++ /dev/null Thu Jan 01 00:00:00 1970 +0000 @@ -1,277 +0,0 @@ -#!/usr/bin/perl -use strict; -use warnings; - -my $read1_row = $ARGV[0]; -my $read2_row = $ARGV[1]; - -my $read1_trimmed = $ARGV[2]; -my $read2_trimmed = $ARGV[3]; - -my $read1_trimmed_noN = $ARGV[4]; -my $read2_trimmed_noN = $ARGV[5]; - -my $sam_row = $ARGV[6]; - -my $sam_filtered = $ARGV[7]; - -my $mpileup_row = $ARGV[8]; - -my $mpileup_filtered = $ARGV[9]; - -my $list_filtered = $ARGV[10]; - -my $blast_filtered = $ARGV[11]; - - - - -open(INR1R, $read1_row) or die ("Can't open $read1_row\n"); -my $nbread=0; -my $nbbase =0; -while (my $line1=){ - my $line2 = ; - my $line3 = ; - my $line4 = ; - if ($line1 =~ /^@/){ - $nbread++; - if ($line2=~/([ATGCNX]+)/i){ - $nbbase += length($1); - } - } -} -print "Row Reads 1\t\tNumber of reads : ",$nbread,"\tnumber of bases : ",$nbbase,"\n"; -close (INR1R); - -open(INR2R, $read2_row) or die ("Can't open $read2_row\n"); -$nbread=0; -$nbbase =0; -while (my $line1=){ - my $line2 = ; - my $line3 = ; - my $line4 = ; - if ($line1 =~ /^@/){ - $nbread++; - if ($line2=~/([ATGCNX]+)/i){ - $nbbase += length($1); - } - } -} -print "Row Reads 2\t\tNumber of reads : ",$nbread,"\tnumber of bases : ",$nbbase,"\n"; -close (INR2R); - -open(INR1T, $read1_trimmed) or die ("Can't open $read1_trimmed\n"); -$nbread=0; -$nbbase =0; -while (my $line1=){ - my $line2 = ; - my $line3 = ; - my $line4 = ; - if ($line1 =~ /^@/){ - $nbread++; - if ($line2=~/([ATGCNX]+)/i){ - $nbbase += length($1); - } - else { - print STDERR "$line1\n$line2\n"; - } - } -} -print "Trimmed Reads 1\t\tNumber of reads : ",$nbread,"\tnumber of bases : ",$nbbase,"\n"; -close (INR1T); - -open(INR2T, $read2_trimmed) or die ("Can't open $read2_trimmed\n"); -$nbread=0; -$nbbase =0; -while (my $line1=){ - my $line2 = ; - my $line3 = ; - my $line4 = ; - if ($line1 =~ /^@/){ - $nbread++; - if ($line2=~/([ATGCNX]+)/i){ - $nbbase += length($1); - } - else { - print STDERR "$line1\n$line2\n"; - } - } -} -print "Trimmed Reads 2\t\tNumber of reads : ",$nbread,"\tnumber of bases : ",$nbbase,"\n"; -close (INR2T); - -open(INR1N, $read1_trimmed_noN) or die ("Can't open $read1_trimmed_noN\n"); -$nbread=0; -$nbbase =0; -while (my $line1=){ - my $line2 = ; - my $line3 = ; - my $line4 = ; - if ($line1 =~ /^@/){ - $nbread++; - if ($line2=~/([ATGCNX]+)/i){ - $nbbase += length($1); - } - else { - print STDERR "$line1\n$line2\n"; - } - } -} -print "Trimmed (no N) Reads 1\tNumber of reads : ",$nbread,"\tnumber of bases : ",$nbbase,"\n"; -close (INR1N); - -open(INR2N, $read2_trimmed_noN) or die ("Can't open $read2_trimmed_noN\n"); -$nbread=0; -$nbbase =0; -while (my $line1=){ - my $line2 = ; - my $line3 = ; - my $line4 = ; - if ($line1 =~ /^@/){ - $nbread++; - if ($line2=~/([ATGCNX]+)/i){ - $nbbase += length($1); - } - else { - print STDERR "$line1\n$line2\n"; - } - } -} -print "Trimmed (no N) Reads 2\tNumber of reads : ",$nbread,"\tnumber of bases : ",$nbbase,"\n"; -close (INR2N); - -print "\nSAM row\n"; -open(SAM, $sam_row) or die ("Can't open $sam_row\n"); -my %bitscore; -while (my $line=){ - if (($line !~ /^\@SQ/)&&($line !~ /^\@PG/)){ - my @fields = split(/\s+/,$line); - my $bit = $fields[1]; - if ($bitscore{$bit}){ - $bitscore{$bit}++; - } - else { - $bitscore{$bit}=1; - } - } -} - -print "bitscore\t"; -foreach my $key (sort {$bitscore{$b} <=> $bitscore{$a}} keys %bitscore) { - print $key,"\t*\t"; -} -print "\n"; - -print " number \t"; -foreach my $key (sort {$bitscore{$b} <=> $bitscore{$a}} keys %bitscore) { - print $bitscore{$key},"\t*\t"; -} -print "\n"; -close (SAM); - -print "\nSAM filtered\n"; -open(SAMF, $sam_filtered) or die ("Can't open $sam_filtered\n"); -undef %bitscore; -while (my $line=){ - if (($line !~ /^\@SQ/)&&($line !~ /^\@PG/)){ - my @fields = split(/\s+/,$line); - my $bit = $fields[1]; - if ($bitscore{$bit}){ - $bitscore{$bit}++; - } - else { - $bitscore{$bit}=1; - } - } -} - -print "bitscore\t"; -foreach my $key (sort {$bitscore{$b} <=> $bitscore{$a}} keys %bitscore) { - print $key,"\t*\t"; -} -print "\n"; - -print " number \t"; -foreach my $key (sort {$bitscore{$b} <=> $bitscore{$a}} keys %bitscore) { - print $bitscore{$key},"\t*\t"; -} -print "\n"; -close (SAMF); - -print "\nMPILEUP row\n"; -open(MPR, $mpileup_row) or die ("Can't open $mpileup_row\n"); -my $basecovered=0; -my $nbvariant=0; -while (my $line=){ - my @fields = split(/\s+/,$line); - if ($#fields >= 4){ - $basecovered++; - my $match = $fields[4]; - $match =~ s/\$//g; #the read start at this position - $match =~ s/\^.//g; #the read end at this position followed by quality char - if ($match =~/[ACGTNacgtn]+/){ - $nbvariant++; - } - } - else { - #print STDERR "Erreur : $line\n"; - } -} - -print "Base covered :\t$basecovered\n"; -print "Variant detected :\t$nbvariant\n"; -close (MPR); - -print "\nMPILEUP filered\n"; -open(MPF, $mpileup_filtered) or die ("Can't open $mpileup_filtered\n"); - -$basecovered=0; -$nbvariant=0; -while (my $line=){ - my @fields = split(/\s+/,$line); - if ($#fields >= 4){ - $basecovered++; - my $match = $fields[4]; - $match =~ s/\$//g; #the read start at this position - $match =~ s/\^.//g; #the read end at this position followed by quality char - if ($match =~/[ACGTNacgtn]+/){ - $nbvariant++; - } - } - else { - #print STDERR "Erreur : $line\n"; - } -} - -print "Variant selected :\t$nbvariant\n"; -close (MPF); - -print "\nMPILEUP filered without dubious position\n"; -open(LF, $list_filtered) or die ("Can't open $list_filtered\n"); -$nbvariant=0; -while (my $line=){ - $nbvariant++; -} - -print "Variant selected :\t$nbvariant\n"; -close (LF); - -print "\nMPILEUP filered without dubious position and BLAST\n"; -open(BF, $blast_filtered) or die ("Can't open $blast_filtered\n"); -$nbvariant=0; -while (my $line=){ - $nbvariant++; -} - -print "Variant selected :\t$nbvariant\n"; -close (BF); - - - - - - - - - - diff -r edddaa8ab855 -r e8e6b962c1f2 rapsodyn/rapsodyn_stats.xml --- a/rapsodyn/rapsodyn_stats.xml Thu Aug 21 08:39:15 2014 -0400 +++ /dev/null Thu Jan 01 00:00:00 1970 +0000 @@ -1,28 +0,0 @@ - -Merge best results from Blast - - rapsodyn_stats.pl $input_read1_row_file $input_read2_row_file $input_read1_trimmed_file $input_read2_trimmed_file $input_read1_trimmednoN_file $input_read2_trimmednoN_file $input_sam_row_file $input_sam_filtered_file $input_mpileup_row_file $input_mpileup_filtered_file $input_list_filtered_file $input_blast_filtered_file > $output_file - - - - - - - - - - - - - - - - - - - - - - - - diff -r edddaa8ab855 -r e8e6b962c1f2 rapsodyn/rapsodyn_stats_filtering.pl --- a/rapsodyn/rapsodyn_stats_filtering.pl Thu Aug 21 08:39:15 2014 -0400 +++ /dev/null Thu Jan 01 00:00:00 1970 +0000 @@ -1,90 +0,0 @@ -#!/usr/bin/perl -use strict; -use warnings; - -my $mpileup_row = $ARGV[0]; - -my $mpileup_filtered = $ARGV[1]; - -my $list_filtered = $ARGV[2]; - -my $blast_filtered = $ARGV[3]; - - -print "\nMPILEUP row\n"; -open(MPR, $mpileup_row) or die ("Can't open $mpileup_row\n"); -my $basecovered=0; -my $nbvariant=0; -while (my $line=){ - my @fields = split(/\s+/,$line); - if ($#fields >= 4){ - $basecovered++; - my $match = $fields[4]; - $match =~ s/\$//g; #the read start at this position - $match =~ s/\^.//g; #the read end at this position followed by quality char - if ($match =~/[ACGTNacgtn]+/){ - $nbvariant++; - } - } - else { - #print STDERR "Erreur : $line\n"; - } -} - -print "Base covered :\t$basecovered\n"; -print "Variant detected :\t$nbvariant\n"; -close (MPR); - -print "\nMPILEUP filered\n"; -open(MPF, $mpileup_filtered) or die ("Can't open $mpileup_filtered\n"); - -$basecovered=0; -$nbvariant=0; -while (my $line=){ - my @fields = split(/\s+/,$line); - if ($#fields >= 4){ - $basecovered++; - my $match = $fields[4]; - $match =~ s/\$//g; #the read start at this position - $match =~ s/\^.//g; #the read end at this position followed by quality char - if ($match =~/[ACGTNacgtn]+/){ - $nbvariant++; - } - } - else { - #print STDERR "Erreur : $line\n"; - } -} - -print "Variant selected :\t$nbvariant\n"; -close (MPF); - -print "\nMPILEUP filered without dubious position\n"; -open(LF, $list_filtered) or die ("Can't open $list_filtered\n"); -$nbvariant=0; -while (my $line=){ - $nbvariant++; -} - -print "Variant selected :\t$nbvariant\n"; -close (LF); - -print "\nMPILEUP filered without dubious position and BLAST\n"; -open(BF, $blast_filtered) or die ("Can't open $blast_filtered\n"); -$nbvariant=0; -while (my $line=){ - $nbvariant++; -} - -print "Variant selected :\t$nbvariant\n"; -close (BF); - - - - - - - - - - diff -r edddaa8ab855 -r e8e6b962c1f2 rapsodyn/rapsodyn_stats_filtering.xml --- a/rapsodyn/rapsodyn_stats_filtering.xml Thu Aug 21 08:39:15 2014 -0400 +++ /dev/null Thu Jan 01 00:00:00 1970 +0000 @@ -1,20 +0,0 @@ - -Merge best results from Blast - - rapsodyn_stats_filtering.pl $input_mpileup_row_file $input_mpileup_filtered_file $input_list_filtered_file $input_blast_filtered_file > $output_file - - - - - - - - - - - - - - - - diff -r edddaa8ab855 -r e8e6b962c1f2 rapsodyn/rapsodyn_stats_mapping.pl --- a/rapsodyn/rapsodyn_stats_mapping.pl Thu Aug 21 08:39:15 2014 -0400 +++ /dev/null Thu Jan 01 00:00:00 1970 +0000 @@ -1,193 +0,0 @@ -#!/usr/bin/perl -use strict; -use warnings; - -my $read1_row = $ARGV[0]; -my $read2_row = $ARGV[1]; - -my $read1_trimmed = $ARGV[2]; -my $read2_trimmed = $ARGV[3]; - -my $read1_trimmed_noN = $ARGV[4]; -my $read2_trimmed_noN = $ARGV[5]; - -my $sam_row = $ARGV[6]; - -my $sam_filtered = $ARGV[7]; - - - - - -open(INR1R, $read1_row) or die ("Can't open $read1_row\n"); -my $nbread=0; -my $nbbase =0; -while (my $line1=){ - my $line2 = ; - my $line3 = ; - my $line4 = ; - if ($line1 =~ /^@/){ - $nbread++; - if ($line2=~/([ATGCNX]+)/i){ - $nbbase += length($1); - } - } -} -print "Row Reads 1\t\tNumber of reads : ",$nbread,"\tnumber of bases : ",$nbbase,"\n"; -close (INR1R); - -open(INR2R, $read2_row) or die ("Can't open $read2_row\n"); -$nbread=0; -$nbbase =0; -while (my $line1=){ - my $line2 = ; - my $line3 = ; - my $line4 = ; - if ($line1 =~ /^@/){ - $nbread++; - if ($line2=~/([ATGCNX]+)/i){ - $nbbase += length($1); - } - } -} -print "Row Reads 2\t\tNumber of reads : ",$nbread,"\tnumber of bases : ",$nbbase,"\n"; -close (INR2R); - -open(INR1T, $read1_trimmed) or die ("Can't open $read1_trimmed\n"); -$nbread=0; -$nbbase =0; -while (my $line1=){ - my $line2 = ; - my $line3 = ; - my $line4 = ; - if ($line1 =~ /^@/){ - $nbread++; - if ($line2=~/([ATGCNX]+)/i){ - $nbbase += length($1); - } - else { - print STDERR "$line1\n$line2\n"; - } - } -} -print "Trimmed Reads 1\t\tNumber of reads : ",$nbread,"\tnumber of bases : ",$nbbase,"\n"; -close (INR1T); - -open(INR2T, $read2_trimmed) or die ("Can't open $read2_trimmed\n"); -$nbread=0; -$nbbase =0; -while (my $line1=){ - my $line2 = ; - my $line3 = ; - my $line4 = ; - if ($line1 =~ /^@/){ - $nbread++; - if ($line2=~/([ATGCNX]+)/i){ - $nbbase += length($1); - } - else { - print STDERR "$line1\n$line2\n"; - } - } -} -print "Trimmed Reads 2\t\tNumber of reads : ",$nbread,"\tnumber of bases : ",$nbbase,"\n"; -close (INR2T); - -open(INR1N, $read1_trimmed_noN) or die ("Can't open $read1_trimmed_noN\n"); -$nbread=0; -$nbbase =0; -while (my $line1=){ - my $line2 = ; - my $line3 = ; - my $line4 = ; - if ($line1 =~ /^@/){ - $nbread++; - if ($line2=~/([ATGCNX]+)/i){ - $nbbase += length($1); - } - else { - print STDERR "$line1\n$line2\n"; - } - } -} -print "Trimmed (no N) Reads 1\tNumber of reads : ",$nbread,"\tnumber of bases : ",$nbbase,"\n"; -close (INR1N); - -open(INR2N, $read2_trimmed_noN) or die ("Can't open $read2_trimmed_noN\n"); -$nbread=0; -$nbbase =0; -while (my $line1=){ - my $line2 = ; - my $line3 = ; - my $line4 = ; - if ($line1 =~ /^@/){ - $nbread++; - if ($line2=~/([ATGCNX]+)/i){ - $nbbase += length($1); - } - else { - print STDERR "$line1\n$line2\n"; - } - } -} -print "Trimmed (no N) Reads 2\tNumber of reads : ",$nbread,"\tnumber of bases : ",$nbbase,"\n"; -close (INR2N); - -print "\nSAM row\n"; -open(SAM, $sam_row) or die ("Can't open $sam_row\n"); -my %bitscore; -while (my $line=){ - if (($line !~ /^\@SQ/)&&($line !~ /^\@PG/)){ - my @fields = split(/\s+/,$line); - my $bit = $fields[1]; - if ($bitscore{$bit}){ - $bitscore{$bit}++; - } - else { - $bitscore{$bit}=1; - } - } -} - -print "bitscore\t"; -foreach my $key (sort {$bitscore{$b} <=> $bitscore{$a}} keys %bitscore) { - print $key,"\t*\t"; -} -print "\n"; - -print " number \t"; -foreach my $key (sort {$bitscore{$b} <=> $bitscore{$a}} keys %bitscore) { - print $bitscore{$key},"\t*\t"; -} -print "\n"; -close (SAM); - -print "\nSAM filtered\n"; -open(SAMF, $sam_filtered) or die ("Can't open $sam_filtered\n"); -undef %bitscore; -while (my $line=){ - if (($line !~ /^\@SQ/)&&($line !~ /^\@PG/)){ - my @fields = split(/\s+/,$line); - my $bit = $fields[1]; - if ($bitscore{$bit}){ - $bitscore{$bit}++; - } - else { - $bitscore{$bit}=1; - } - } -} - -print "bitscore\t"; -foreach my $key (sort {$bitscore{$b} <=> $bitscore{$a}} keys %bitscore) { - print $key,"\t*\t"; -} -print "\n"; - -print " number \t"; -foreach my $key (sort {$bitscore{$b} <=> $bitscore{$a}} keys %bitscore) { - print $bitscore{$key},"\t*\t"; -} -print "\n"; -close (SAMF); - diff -r edddaa8ab855 -r e8e6b962c1f2 rapsodyn/rapsodyn_stats_mapping.xml --- a/rapsodyn/rapsodyn_stats_mapping.xml Thu Aug 21 08:39:15 2014 -0400 +++ /dev/null Thu Jan 01 00:00:00 1970 +0000 @@ -1,24 +0,0 @@ - -Stats on the mapping - - rapsodyn_stats_mapping.pl $input_read1_row_file $input_read2_row_file $input_read1_trimmed_file $input_read2_trimmed_file $input_read1_trimmednoN_file $input_read2_trimmednoN_file $input_sam_row_file $input_sam_filtered_file > $output_file - - - - - - - - - - - - - - - - - - - - diff -r edddaa8ab855 -r e8e6b962c1f2 rapsodyn/repository_dependencies.xml --- a/rapsodyn/repository_dependencies.xml Thu Aug 21 08:39:15 2014 -0400 +++ b/rapsodyn/repository_dependencies.xml Fri Sep 05 06:12:10 2014 -0400 @@ -1,10 +1,8 @@ - - - - - - - + + + + +