Mercurial > repos > mcharles > rapsosnp
changeset 8:d857538d9fea draft
Uploaded
| author | mcharles |
|---|---|
| date | Fri, 10 Oct 2014 07:05:36 -0400 |
| parents | 3f7b0788a1c4 |
| children | 0e7c6fe60646 |
| files | rapsodyn/Galaxy-Workflow-rapsosnp_v1.20.ga rapsodyn/Galaxy-Workflow-rapsosnp_v1.20_-_10x_-_sam.ga rapsodyn/Galaxy-Workflow-rapsosnp_v1.20_-_1x_-_sam.ga rapsodyn/PrepareFastqLight.pl rapsodyn/PrepareFastqLight.pl~ rapsodyn/PrepareFastqLight.xml rapsodyn/PrepareFastqLight.xml~ rapsodyn/listfiltering.pl rapsodyn/listfiltering.xml |
| diffstat | 9 files changed, 884 insertions(+), 1200 deletions(-) [+] |
line wrap: on
line diff
--- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/rapsodyn/Galaxy-Workflow-rapsosnp_v1.20.ga Fri Oct 10 07:05:36 2014 -0400 @@ -0,0 +1,604 @@ +{ + "a_galaxy_workflow": "true", + "annotation": "", + "format-version": "0.1", + "name": "rapsosnp v1.20", + "steps": { + "0": { + "annotation": "", + "id": 0, + "input_connections": {}, + "inputs": [ + { + "description": "", + "name": "ASSEMBLY" + } + ], + "name": "Input dataset", + "outputs": [], + "position": { + "left": 200, + "top": 221 + }, + "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": 306.76666259765625 + }, + "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": 392.5333251953125 + }, + "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": 478.29998779296875 + }, + "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": 427.76666259765625, + "top": 221 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "fastaGroomerForMakeBlastdb", + "tool_state": "{\"input_fasta\": \"null\", \"__rerun_remap_job_id__\": null, \"__page__\": 0}", + "tool_version": "1.00", + "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" + }, + { + "name": "log_file", + "type": "txt" + } + ], + "position": { + "left": 427.76666259765625, + "top": 351.76666259765625 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "PrepareFastqLight", + "tool_state": "{\"__page__\": 0, \"quality_type\": \"\\\"auto\\\"\", \"min_length\": \"\\\"30\\\"\", \"__rerun_remap_job_id__\": null, \"min_quality\": \"\\\"30\\\"\", \"input_read1_file\": \"null\", \"input_read2_file\": \"null\"}", + "tool_version": "1.03", + "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": 755.5333251953125, + "top": 221 + }, + "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": 755.5333251953125, + "top": 378.76666259765625 + }, + "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\\\"\"}", + "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_sam_file", + "type": "sam" + }, + { + "name": "log_file", + "type": "txt" + } + ], + "position": { + "left": 1083.300048828125, + "top": 221 + }, + "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": "1.01", + "type": "tool", + "user_outputs": [] + }, + "9": { + "annotation": "", + "id": 9, + "input_connections": { + "source|input1": { + "id": 8, + "output_name": "output_sam_file" + }, + "source|ref_file": { + "id": 0, + "output_name": "output" + } + }, + "inputs": [], + "name": "SAM-to-BAM", + "outputs": [ + { + "name": "output1", + "type": "bam" + } + ], + "position": { + "left": 1411.066650390625, + "top": 221 + }, + "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, \"__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": 1638.8333740234375, + "top": 221 + }, + "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, \"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\\\": \\\"0\\\", \\\"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\\\"}\", \"__rerun_remap_job_id__\": null, \"genotype_likelihood_computation_type\": \"{\\\"genotype_likelihood_computation_type_selector\\\": \\\"do_not_perform_genotype_likelihood_computation\\\", \\\"__current_case__\\\": 1}\", \"reference_source\": \"{\\\"ref_file\\\": null, \\\"reference_source_selector\\\": \\\"history\\\", \\\"input_bams\\\": [{\\\"__index__\\\": 0, \\\"input_bam\\\": null}], \\\"__current_case__\\\": 1}\"}", + "tool_version": "0.0.3", + "type": "tool", + "user_outputs": [] + }, + "11": { + "annotation": "", + "id": 11, + "input_connections": { + "input_pileup_file": { + "id": 10, + "output_name": "output_mpileup" + } + }, + "inputs": [], + "name": "PileupVariant", + "outputs": [ + { + "name": "output_pileup_file", + "type": "pileup" + }, + { + "name": "log_file", + "type": "txt" + } + ], + "position": { + "left": 1964.60009765625, + "top": 221 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "PileupVariant", + "tool_state": "{\"__page__\": 0, \"__rerun_remap_job_id__\": null, \"input_pileup_file\": \"null\"}", + "tool_version": "1.01", + "type": "tool", + "user_outputs": [] + }, + "12": { + "annotation": "", + "id": 12, + "input_connections": { + "input_list1": { + "id": 11, + "output_name": "output_pileup_file" + }, + "input_list2": { + "id": 3, + "output_name": "output" + } + }, + "inputs": [], + "name": "listfiltering", + "outputs": [ + { + "name": "output_file", + "type": "pileup" + }, + { + "name": "log_file", + "type": "txt" + } + ], + "position": { + "left": 2292.36669921875, + "top": 221 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "listfiltering", + "tool_state": "{\"__page__\": 0, \"input_list2\": \"null\", \"input_list1\": \"null\", \"__rerun_remap_job_id__\": null, \"nb_col\": \"\\\"2\\\"\", \"type\": \"\\\"specific\\\"\"}", + "tool_version": "1.01", + "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": 2561.13330078125, + "top": 221 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "extractseq", + "tool_state": "{\"__page__\": 0, \"input_variant_file\": \"null\", \"__rerun_remap_job_id__\": null, \"window_length\": \"\\\"50\\\"\", \"input_assembly_file\": \"null\"}", + "tool_version": "1.01", + "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": 2888.89990234375, + "top": 221 + }, + "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\\\": \\\"8\\\"}\", \"__rerun_remap_job_id__\": null, \"blast_type\": \"\\\"megablast\\\"\", \"db_opts\": \"{\\\"db_opts_selector\\\": \\\"histdb\\\", \\\"subject\\\": \\\"\\\", \\\"histdb\\\": null, \\\"__current_case__\\\": 1, \\\"database\\\": \\\"\\\"}\", \"output\": \"{\\\"out_format\\\": \\\"cols\\\", \\\"std_cols\\\": [\\\"qseqid\\\", \\\"qstart\\\", \\\"qend\\\"], \\\"ids_cols\\\": null, \\\"tax_cols\\\": null, \\\"__current_case__\\\": 2, \\\"misc_cols\\\": null, \\\"ext_cols\\\": [\\\"qseq\\\", \\\"sseq\\\"]}\", \"query\": \"null\"}", + "tool_version": "0.1.00", + "type": "tool", + "user_outputs": [] + }, + "15": { + "annotation": "", + "id": 15, + "input_connections": { + "input_blast_file": { + "id": 14, + "output_name": "output1" + }, + "input_variant_file": { + "id": 12, + "output_name": "output_file" + } + }, + "inputs": [], + "name": "ParseBlastForUniqueMatch", + "outputs": [ + { + "name": "output_variant_file", + "type": "pileup" + }, + { + "name": "log_file", + "type": "txt" + } + ], + "position": { + "left": 3190.666748046875, + "top": 221 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "ParseBlastForUniqueMatch", + "tool_state": "{\"input_variant_file\": \"null\", \"__page__\": 0, \"__rerun_remap_job_id__\": null, \"input_blast_file\": \"null\", \"window_length\": \"\\\"50\\\"\", \"nb_mismatch_max\": \"\\\"3\\\"\"}", + "tool_version": "1.01", + "type": "tool", + "user_outputs": [] + }, + "16": { + "annotation": "", + "id": 16, + "input_connections": { + "input_file": { + "id": 15, + "output_name": "output_variant_file" + } + }, + "inputs": [], + "name": "mpileupfilterandstat", + "outputs": [ + { + "name": "output_file", + "type": "pileup" + }, + { + "name": "log_file", + "type": "txt" + } + ], + "position": { + "left": 3518.433349609375, + "top": 221 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "mpileupfilterandstat", + "tool_state": "{\"__page__\": 0, \"stat\": \"{\\\"stat_dist_step\\\": \\\"50\\\", \\\"stat_min_depth_step\\\": \\\"4\\\", \\\"stat_max_depth_max\\\": \\\"200\\\", \\\"stat_min_depth_max\\\": \\\"20\\\", \\\"stat_freq_max\\\": \\\"1.0\\\", \\\"stat_min_depth_min\\\": \\\"4\\\", \\\"stat_max_depth_min\\\": \\\"200\\\", \\\"do_stat\\\": \\\"YES\\\", \\\"stat_dist_min\\\": \\\"0\\\", \\\"__current_case__\\\": 0, \\\"stat_max_depth_step\\\": \\\"100\\\", \\\"stat_dist_max\\\": \\\"50\\\", \\\"stat_freq_step\\\": \\\"0.2\\\", \\\"stat_freq_min\\\": \\\"0.8\\\"}\", \"input_file\": \"null\", \"min_frequency\": \"\\\"0.0\\\"\", \"min_depth\": \"\\\"1\\\"\", \"__rerun_remap_job_id__\": null, \"min_forward_and_reverse\": \"\\\"0\\\"\", \"max_depth\": \"\\\"500\\\"\", \"min_distance\": \"\\\"0\\\"\"}", + "tool_version": "1.00", + "type": "tool", + "user_outputs": [] + }, + "17": { + "annotation": "", + "id": 17, + "input_connections": { + "input_files": [ + { + "id": 5, + "output_name": "log_file" + }, + { + "id": 8, + "output_name": "log_file" + }, + { + "id": 11, + "output_name": "log_file" + }, + { + "id": 12, + "output_name": "log_file" + }, + { + "id": 15, + "output_name": "log_file" + }, + { + "id": 16, + "output_name": "log_file" + } + ] + }, + "inputs": [], + "name": "MergeMultiFile", + "outputs": [ + { + "name": "output_file", + "type": "txt" + } + ], + "position": { + "left": 3846.2001953125, + "top": 221 + }, + "post_job_actions": { + "RenameDatasetActionoutput_file": { + "action_arguments": { + "newname": "LOGS" + }, + "action_type": "RenameDatasetAction", + "output_name": "output_file" + } + }, + "tool_errors": null, + "tool_id": "MergeMultiFile", + "tool_state": "{\"out_format\": \"\\\"txt\\\"\", \"__page__\": 0, \"__rerun_remap_job_id__\": null, \"input_files\": \"null\"}", + "tool_version": "1.00", + "type": "tool", + "user_outputs": [] + } + } +} \ No newline at end of file
--- a/rapsodyn/Galaxy-Workflow-rapsosnp_v1.20_-_10x_-_sam.ga Tue Oct 07 10:34:34 2014 -0400 +++ b/rapsodyn/Galaxy-Workflow-rapsosnp_v1.20_-_10x_-_sam.ga Fri Oct 10 07:05:36 2014 -0400 @@ -2,7 +2,7 @@ "a_galaxy_workflow": "true", "annotation": "", "format-version": "0.1", - "name": "rapsosnp v1.20 - 1x - sam", + "name": "rapsosnp v1.20 - 10x - sam", "steps": { "0": { "annotation": "", @@ -285,14 +285,14 @@ } ], "position": { - "left": 936.93310546875, + "left": 936.933349609375, "top": 338.43333435058594 }, "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": "1.01", + "tool_state": "{\"__page__\": 0, \"quality_type\": \"\\\"auto\\\"\", \"min_length\": \"\\\"30\\\"\", \"__rerun_remap_job_id__\": null, \"min_quality\": \"\\\"30\\\"\", \"input_read1_file\": \"null\", \"input_read2_file\": \"null\"}", + "tool_version": "1.03", "type": "tool", "user_outputs": [] }, @@ -326,14 +326,14 @@ } ], "position": { - "left": 943.93310546875, + "left": 943.933349609375, "top": 585.4333343505859 }, "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": "1.01", + "tool_state": "{\"__page__\": 0, \"quality_type\": \"\\\"auto\\\"\", \"min_length\": \"\\\"30\\\"\", \"__rerun_remap_job_id__\": null, \"min_quality\": \"\\\"30\\\"\", \"input_read1_file\": \"null\", \"input_read2_file\": \"null\"}", + "tool_version": "1.03", "type": "tool", "user_outputs": [] }, @@ -367,14 +367,14 @@ } ], "position": { - "left": 947.93310546875, + "left": 947.933349609375, "top": 812.433349609375 }, "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": "1.01", + "tool_state": "{\"__page__\": 0, \"quality_type\": \"\\\"auto\\\"\", \"min_length\": \"\\\"30\\\"\", \"__rerun_remap_job_id__\": null, \"min_quality\": \"\\\"30\\\"\", \"input_read1_file\": \"null\", \"input_read2_file\": \"null\"}", + "tool_version": "1.03", "type": "tool", "user_outputs": [] }, @@ -408,14 +408,14 @@ } ], "position": { - "left": 1267.93310546875, + "left": 1267.933349609375, "top": 343.43333435058594 }, "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": "1.01", + "tool_state": "{\"__page__\": 0, \"quality_type\": \"\\\"auto\\\"\", \"min_length\": \"\\\"30\\\"\", \"__rerun_remap_job_id__\": null, \"min_quality\": \"\\\"30\\\"\", \"input_read1_file\": \"null\", \"input_read2_file\": \"null\"}", + "tool_version": "1.03", "type": "tool", "user_outputs": [] }, @@ -449,14 +449,14 @@ } ], "position": { - "left": 1261.93310546875, + "left": 1261.933349609375, "top": 587.4333343505859 }, "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": "1.01", + "tool_state": "{\"__page__\": 0, \"quality_type\": \"\\\"auto\\\"\", \"min_length\": \"\\\"30\\\"\", \"__rerun_remap_job_id__\": null, \"min_quality\": \"\\\"30\\\"\", \"input_read1_file\": \"null\", \"input_read2_file\": \"null\"}", + "tool_version": "1.03", "type": "tool", "user_outputs": [] }, @@ -490,14 +490,14 @@ } ], "position": { - "left": 954.93310546875, + "left": 954.933349609375, "top": 1044.433349609375 }, "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": "1.01", + "tool_state": "{\"__page__\": 0, \"quality_type\": \"\\\"auto\\\"\", \"min_length\": \"\\\"30\\\"\", \"__rerun_remap_job_id__\": null, \"min_quality\": \"\\\"30\\\"\", \"input_read1_file\": \"null\", \"input_read2_file\": \"null\"}", + "tool_version": "1.03", "type": "tool", "user_outputs": [] }, @@ -531,14 +531,14 @@ } ], "position": { - "left": 1265.93310546875, + "left": 1265.933349609375, "top": 815.433349609375 }, "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": "1.01", + "tool_state": "{\"__page__\": 0, \"quality_type\": \"\\\"auto\\\"\", \"min_length\": \"\\\"30\\\"\", \"__rerun_remap_job_id__\": null, \"min_quality\": \"\\\"30\\\"\", \"input_read1_file\": \"null\", \"input_read2_file\": \"null\"}", + "tool_version": "1.03", "type": "tool", "user_outputs": [] }, @@ -572,14 +572,14 @@ } ], "position": { - "left": 953.93310546875, + "left": 953.933349609375, "top": 1282.433349609375 }, "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": "1.01", + "tool_state": "{\"__page__\": 0, \"quality_type\": \"\\\"auto\\\"\", \"min_length\": \"\\\"30\\\"\", \"__rerun_remap_job_id__\": null, \"min_quality\": \"\\\"30\\\"\", \"input_read1_file\": \"null\", \"input_read2_file\": \"null\"}", + "tool_version": "1.03", "type": "tool", "user_outputs": [] }, @@ -613,14 +613,14 @@ } ], "position": { - "left": 1265.93310546875, + "left": 1265.933349609375, "top": 1043.433349609375 }, "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": "1.01", + "tool_state": "{\"__page__\": 0, \"quality_type\": \"\\\"auto\\\"\", \"min_length\": \"\\\"30\\\"\", \"__rerun_remap_job_id__\": null, \"min_quality\": \"\\\"30\\\"\", \"input_read1_file\": \"null\", \"input_read2_file\": \"null\"}", + "tool_version": "1.03", "type": "tool", "user_outputs": [] }, @@ -654,14 +654,14 @@ } ], "position": { - "left": 1270.93310546875, + "left": 1270.933349609375, "top": 1281.433349609375 }, "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": "1.01", + "tool_state": "{\"__page__\": 0, \"quality_type\": \"\\\"auto\\\"\", \"min_length\": \"\\\"30\\\"\", \"__rerun_remap_job_id__\": null, \"min_quality\": \"\\\"30\\\"\", \"input_read1_file\": \"null\", \"input_read2_file\": \"null\"}", + "tool_version": "1.03", "type": "tool", "user_outputs": [] }, @@ -1400,15 +1400,27 @@ "input_connections": { "input_files": [ { - "id": 37, + "id": 28, + "output_name": "output_sam_file" + }, + { + "id": 29, "output_name": "output_sam_file" }, { - "id": 36, + "id": 30, "output_name": "output_sam_file" }, { - "id": 35, + "id": 31, + "output_name": "output_sam_file" + }, + { + "id": 32, + "output_name": "output_sam_file" + }, + { + "id": 33, "output_name": "output_sam_file" }, { @@ -1416,27 +1428,15 @@ "output_name": "output_sam_file" }, { - "id": 33, - "output_name": "output_sam_file" - }, - { - "id": 32, + "id": 35, "output_name": "output_sam_file" }, { - "id": 31, + "id": 36, "output_name": "output_sam_file" }, { - "id": 30, - "output_name": "output_sam_file" - }, - { - "id": 29, - "output_name": "output_sam_file" - }, - { - "id": 28, + "id": 37, "output_name": "output_sam_file" } ] @@ -2702,15 +2702,27 @@ "input_connections": { "input_files": [ { - "id": 64, + "id": 69, + "output_name": "output_variant_file" + }, + { + "id": 73, "output_name": "output_variant_file" }, { - "id": 65, + "id": 72, "output_name": "output_variant_file" }, { - "id": 66, + "id": 71, + "output_name": "output_variant_file" + }, + { + "id": 70, + "output_name": "output_variant_file" + }, + { + "id": 68, "output_name": "output_variant_file" }, { @@ -2718,27 +2730,15 @@ "output_name": "output_variant_file" }, { - "id": 68, - "output_name": "output_variant_file" - }, - { - "id": 70, + "id": 66, "output_name": "output_variant_file" }, { - "id": 71, + "id": 65, "output_name": "output_variant_file" }, { - "id": 72, - "output_name": "output_variant_file" - }, - { - "id": 73, - "output_name": "output_variant_file" - }, - { - "id": 69, + "id": 64, "output_name": "output_variant_file" } ] @@ -2802,63 +2802,71 @@ "input_connections": { "input_files": [ { - "id": 7, + "id": 75, "output_name": "log_file" }, { - "id": 8, + "id": 69, "output_name": "log_file" }, { - "id": 9, + "id": 73, "output_name": "log_file" }, { - "id": 12, + "id": 72, + "output_name": "log_file" + }, + { + "id": 71, "output_name": "log_file" }, { - "id": 14, + "id": 70, "output_name": "log_file" }, { - "id": 10, + "id": 68, "output_name": "log_file" }, { - "id": 11, + "id": 67, "output_name": "log_file" }, { - "id": 13, + "id": 66, "output_name": "log_file" }, { - "id": 15, + "id": 65, "output_name": "log_file" }, { - "id": 16, + "id": 64, + "output_name": "log_file" + }, + { + "id": 42, "output_name": "log_file" }, { - "id": 28, + "id": 41, "output_name": "log_file" }, { - "id": 29, + "id": 37, "output_name": "log_file" }, { - "id": 30, + "id": 36, "output_name": "log_file" }, { - "id": 33, + "id": 34, "output_name": "log_file" }, { - "id": 35, + "id": 32, "output_name": "log_file" }, { @@ -2866,71 +2874,63 @@ "output_name": "log_file" }, { - "id": 32, + "id": 35, "output_name": "log_file" }, { - "id": 34, + "id": 33, "output_name": "log_file" }, { - "id": 36, + "id": 30, "output_name": "log_file" }, { - "id": 37, - "output_name": "log_file" - }, - { - "id": 41, + "id": 29, "output_name": "log_file" }, { - "id": 42, + "id": 28, "output_name": "log_file" }, { - "id": 64, + "id": 16, "output_name": "log_file" }, { - "id": 65, + "id": 15, "output_name": "log_file" }, { - "id": 66, + "id": 13, "output_name": "log_file" }, { - "id": 67, + "id": 11, "output_name": "log_file" }, { - "id": 68, - "output_name": "log_file" - }, - { - "id": 70, + "id": 10, "output_name": "log_file" }, { - "id": 71, + "id": 14, "output_name": "log_file" }, { - "id": 72, + "id": 12, "output_name": "log_file" }, { - "id": 73, + "id": 9, "output_name": "log_file" }, { - "id": 69, + "id": 8, "output_name": "log_file" }, { - "id": 75, + "id": 7, "output_name": "log_file" } ]
--- a/rapsodyn/Galaxy-Workflow-rapsosnp_v1.20_-_1x_-_sam.ga Tue Oct 07 10:34:34 2014 -0400 +++ /dev/null Thu Jan 01 00:00:00 1970 +0000 @@ -1,604 +0,0 @@ -{ - "a_galaxy_workflow": "true", - "annotation": "", - "format-version": "0.1", - "name": "rapsosnp v1.20 - 1x - sam", - "steps": { - "0": { - "annotation": "", - "id": 0, - "input_connections": {}, - "inputs": [ - { - "description": "", - "name": "ASSEMBLY" - } - ], - "name": "Input dataset", - "outputs": [], - "position": { - "left": 200, - "top": 221 - }, - "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": 306.76666259765625 - }, - "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": 392.5333251953125 - }, - "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": 478.29998779296875 - }, - "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": 427.7666015625, - "top": 221 - }, - "post_job_actions": {}, - "tool_errors": null, - "tool_id": "fastaGroomerForMakeBlastdb", - "tool_state": "{\"input_fasta\": \"null\", \"__rerun_remap_job_id__\": null, \"__page__\": 0}", - "tool_version": "1.00", - "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" - }, - { - "name": "log_file", - "type": "txt" - } - ], - "position": { - "left": 427.7666015625, - "top": 351.76666259765625 - }, - "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": "1.01", - "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": 755.5333251953125, - "top": 221 - }, - "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": 755.5333251953125, - "top": 378.76666259765625 - }, - "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\\\"\"}", - "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_sam_file", - "type": "sam" - }, - { - "name": "log_file", - "type": "txt" - } - ], - "position": { - "left": 1083.300048828125, - "top": 221 - }, - "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": "1.01", - "type": "tool", - "user_outputs": [] - }, - "9": { - "annotation": "", - "id": 9, - "input_connections": { - "source|input1": { - "id": 8, - "output_name": "output_sam_file" - }, - "source|ref_file": { - "id": 0, - "output_name": "output" - } - }, - "inputs": [], - "name": "SAM-to-BAM", - "outputs": [ - { - "name": "output1", - "type": "bam" - } - ], - "position": { - "left": 1411.066650390625, - "top": 221 - }, - "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, \"__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": 1638.8333129882812, - "top": 221 - }, - "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, \"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\\\": \\\"0\\\", \\\"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\\\"}\", \"__rerun_remap_job_id__\": null, \"genotype_likelihood_computation_type\": \"{\\\"genotype_likelihood_computation_type_selector\\\": \\\"do_not_perform_genotype_likelihood_computation\\\", \\\"__current_case__\\\": 1}\", \"reference_source\": \"{\\\"ref_file\\\": null, \\\"reference_source_selector\\\": \\\"history\\\", \\\"input_bams\\\": [{\\\"__index__\\\": 0, \\\"input_bam\\\": null}], \\\"__current_case__\\\": 1}\"}", - "tool_version": "0.0.3", - "type": "tool", - "user_outputs": [] - }, - "11": { - "annotation": "", - "id": 11, - "input_connections": { - "input_pileup_file": { - "id": 10, - "output_name": "output_mpileup" - } - }, - "inputs": [], - "name": "PileupVariant", - "outputs": [ - { - "name": "output_pileup_file", - "type": "pileup" - }, - { - "name": "log_file", - "type": "txt" - } - ], - "position": { - "left": 1964.5999755859375, - "top": 221 - }, - "post_job_actions": {}, - "tool_errors": null, - "tool_id": "PileupVariant", - "tool_state": "{\"__page__\": 0, \"__rerun_remap_job_id__\": null, \"input_pileup_file\": \"null\"}", - "tool_version": "1.01", - "type": "tool", - "user_outputs": [] - }, - "12": { - "annotation": "", - "id": 12, - "input_connections": { - "input_list1": { - "id": 11, - "output_name": "output_pileup_file" - }, - "input_list2": { - "id": 3, - "output_name": "output" - } - }, - "inputs": [], - "name": "listfiltering", - "outputs": [ - { - "name": "output_file", - "type": "pileup" - }, - { - "name": "log_file", - "type": "txt" - } - ], - "position": { - "left": 2292.366668701172, - "top": 221 - }, - "post_job_actions": {}, - "tool_errors": null, - "tool_id": "listfiltering", - "tool_state": "{\"__page__\": 0, \"input_list2\": \"null\", \"input_list1\": \"null\", \"__rerun_remap_job_id__\": null, \"nb_col\": \"\\\"2\\\"\", \"type\": \"\\\"specific\\\"\"}", - "tool_version": "1.01", - "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": 2561.133331298828, - "top": 221 - }, - "post_job_actions": {}, - "tool_errors": null, - "tool_id": "extractseq", - "tool_state": "{\"__page__\": 0, \"input_variant_file\": \"null\", \"__rerun_remap_job_id__\": null, \"window_length\": \"\\\"50\\\"\", \"input_assembly_file\": \"null\"}", - "tool_version": "1.01", - "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": 2888.8999938964844, - "top": 221 - }, - "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\\\": \\\"8\\\"}\", \"__rerun_remap_job_id__\": null, \"blast_type\": \"\\\"megablast\\\"\", \"db_opts\": \"{\\\"db_opts_selector\\\": \\\"histdb\\\", \\\"subject\\\": \\\"\\\", \\\"histdb\\\": null, \\\"__current_case__\\\": 1, \\\"database\\\": \\\"\\\"}\", \"output\": \"{\\\"out_format\\\": \\\"cols\\\", \\\"std_cols\\\": [\\\"qseqid\\\", \\\"qstart\\\", \\\"qend\\\"], \\\"ids_cols\\\": null, \\\"tax_cols\\\": null, \\\"__current_case__\\\": 2, \\\"misc_cols\\\": null, \\\"ext_cols\\\": [\\\"qseq\\\", \\\"sseq\\\"]}\", \"query\": \"null\"}", - "tool_version": "0.1.00", - "type": "tool", - "user_outputs": [] - }, - "15": { - "annotation": "", - "id": 15, - "input_connections": { - "input_blast_file": { - "id": 14, - "output_name": "output1" - }, - "input_variant_file": { - "id": 12, - "output_name": "output_file" - } - }, - "inputs": [], - "name": "ParseBlastForUniqueMatch", - "outputs": [ - { - "name": "output_variant_file", - "type": "pileup" - }, - { - "name": "log_file", - "type": "txt" - } - ], - "position": { - "left": 3190.6666870117188, - "top": 221 - }, - "post_job_actions": {}, - "tool_errors": null, - "tool_id": "ParseBlastForUniqueMatch", - "tool_state": "{\"input_variant_file\": \"null\", \"__page__\": 0, \"__rerun_remap_job_id__\": null, \"input_blast_file\": \"null\", \"window_length\": \"\\\"50\\\"\", \"nb_mismatch_max\": \"\\\"3\\\"\"}", - "tool_version": "1.01", - "type": "tool", - "user_outputs": [] - }, - "16": { - "annotation": "", - "id": 16, - "input_connections": { - "input_file": { - "id": 15, - "output_name": "output_variant_file" - } - }, - "inputs": [], - "name": "mpileupfilterandstat", - "outputs": [ - { - "name": "output_file", - "type": "pileup" - }, - { - "name": "log_file", - "type": "txt" - } - ], - "position": { - "left": 3518.433349609375, - "top": 221 - }, - "post_job_actions": {}, - "tool_errors": null, - "tool_id": "mpileupfilterandstat", - "tool_state": "{\"__page__\": 0, \"stat\": \"{\\\"stat_dist_step\\\": \\\"50\\\", \\\"stat_min_depth_step\\\": \\\"4\\\", \\\"stat_max_depth_max\\\": \\\"200\\\", \\\"stat_min_depth_max\\\": \\\"20\\\", \\\"stat_freq_max\\\": \\\"1.0\\\", \\\"stat_min_depth_min\\\": \\\"4\\\", \\\"stat_max_depth_min\\\": \\\"200\\\", \\\"do_stat\\\": \\\"YES\\\", \\\"stat_dist_min\\\": \\\"0\\\", \\\"__current_case__\\\": 0, \\\"stat_max_depth_step\\\": \\\"100\\\", \\\"stat_dist_max\\\": \\\"50\\\", \\\"stat_freq_step\\\": \\\"0.2\\\", \\\"stat_freq_min\\\": \\\"0.8\\\"}\", \"input_file\": \"null\", \"min_frequency\": \"\\\"0.0\\\"\", \"min_depth\": \"\\\"1\\\"\", \"__rerun_remap_job_id__\": null, \"min_forward_and_reverse\": \"\\\"0\\\"\", \"max_depth\": \"\\\"500\\\"\", \"min_distance\": \"\\\"0\\\"\"}", - "tool_version": "1.00", - "type": "tool", - "user_outputs": [] - }, - "17": { - "annotation": "", - "id": 17, - "input_connections": { - "input_files": [ - { - "id": 16, - "output_name": "log_file" - }, - { - "id": 15, - "output_name": "log_file" - }, - { - "id": 12, - "output_name": "log_file" - }, - { - "id": 11, - "output_name": "log_file" - }, - { - "id": 8, - "output_name": "log_file" - }, - { - "id": 5, - "output_name": "log_file" - } - ] - }, - "inputs": [], - "name": "MergeMultiFile", - "outputs": [ - { - "name": "output_file", - "type": "txt" - } - ], - "position": { - "left": 3846.199951171875, - "top": 221 - }, - "post_job_actions": { - "RenameDatasetActionoutput_file": { - "action_arguments": { - "newname": "LOGS" - }, - "action_type": "RenameDatasetAction", - "output_name": "output_file" - } - }, - "tool_errors": null, - "tool_id": "MergeMultiFile", - "tool_state": "{\"out_format\": \"\\\"txt\\\"\", \"__page__\": 0, \"__rerun_remap_job_id__\": null, \"input_files\": \"null\"}", - "tool_version": "1.00", - "type": "tool", - "user_outputs": [] - } - } -} \ No newline at end of file
--- a/rapsodyn/PrepareFastqLight.pl Tue Oct 07 10:34:34 2014 -0400 +++ b/rapsodyn/PrepareFastqLight.pl Fri Oct 10 07:05:36 2014 -0400 @@ -1,4 +1,6 @@ #!/usr/bin/perl +#v1.0.3 support rapsodyn header (.... 1:... / .... 2:...) +#V1.0.2 added auto type detection #V1.0.1 added log, option parameters use strict; use warnings; @@ -41,11 +43,12 @@ my $nb_base_current_t=0; - open(READ1, $read1_file) or die ("Can't open $read1_file\n"); open(READ2, $read2_file) or die ("Can't open $read2_file\n"); open(OUT1, ">$output1_file") or die ("Can't open $output1_file\n"); open(OUT2, ">$output2_file") or die ("Can't open $output2_file\n"); +open (LF,">$log_file") or die("Can't open $log_file\n"); + my $error1=0; my $error2=0; @@ -58,6 +61,59 @@ my $error9=0; my $error10=0; +my $auto_type=""; +my %qual; +if ($TYPE eq "auto"){ + my $compt=0; + open(DETECT, $read1_file) or die ("Can't open $read1_file\n"); + while (my $ligne1_r1 =<DETECT>){ + my $ligne2_r1 =<DETECT>; + my $ligne3_r1 =<DETECT>; + my $ligne4_r1 =<DETECT>; + $compt++; + if ($ligne4_r1 =~ /^(.*)\s*$/i){ + my $qual = $1; + my @q = split(//,$qual); + for (my $i=0;$i<=$#q;$i++){ + my $num = ord($q[$i]); + if ($qual{$num}){ + $qual{$num}++; + } + else { + $qual{$num} = 1; + } + #range sanger / illumina 1.8+ : 33->94 + #range illumina 1.3->1.7 : 64->105 + if ($num > 94){$auto_type = "illumina";last;} + if ($num < 64){$auto_type = "sanger";last;} + } + } + else { + print STDERR "Error in format detection : quality not recognized\n$ligne4_r1"; + exit(0); + } + + if ($auto_type ne ""){ + last; + } + + } + close (DETECT); + if ($auto_type eq ""){ + print STDERR "Error in format detection : type not recognized parsing read1\n"; + foreach my $key (sort {$a <=> $b} keys %qual){ + print "$key\t:\t",$qual{$key},"\n"; + } + exit(0); + } + else { + $TYPE = $auto_type; + } +} + + + + while (my $ligne1_r1 =<READ1>){ my $ligne2_r1 =<READ1>; my $ligne3_r1 =<READ1>; @@ -119,19 +175,19 @@ my $repheader2=""; - if ($ligne1_r1 =~/^\@(.*?)\#/){ + if ($ligne1_r1 =~/^\@(.*?)[\s\/]/){ $header1 = $1; } - if ($ligne3_r1 =~/^\+(.*?)\#/){ + if ($ligne3_r1 =~/^\+(.*?)[\s\/]/){ $repheader1 = $1; } - if ($ligne1_r2 =~/^\@(.*?)\#/){ + if ($ligne1_r2 =~/^\@(.*?)[\s\/]/){ $header2 = $1; } - if ($ligne3_r2 =~/^\+(.*?)\#/){ + if ($ligne3_r2 =~/^\+(.*?)[\s\/]/){ $repheader2 = $1; } #@ 2 sec @@ -298,8 +354,8 @@ close (OUT1); close (OUT2); -open (LF,">$log_file") or die("Can't open $log_file\n"); print LF "\n####\t Fastq preparation \n"; +print LF "Fastq format : $TYPE\n"; print LF "## Before preparation\n"; print LF "#Read1 :\t$nb_read1\t#Base :\t$nb_base_read1\n"; print LF "#Read2 :\t$nb_read2\t#Base :\t$nb_base_read2\n";
--- a/rapsodyn/PrepareFastqLight.pl~ Tue Oct 07 10:34:34 2014 -0400 +++ /dev/null Thu Jan 01 00:00:00 1970 +0000 @@ -1,457 +0,0 @@ -#!/usr/bin/perl -#V1.0.1 added log, option parameters -use strict; -use warnings; -use Getopt::Long; - -my $read1_file; -my $read2_file; -my $log_file; -my $output1_file; -my $output2_file; - -my $TYPE="sanger"; -my $MIN_LENGTH=30; -my $MIN_QUALITY=30; - -my $VERBOSE = "OFF"; - -GetOptions ( -"read1_file=s" => \$read1_file, -"read2_file=s" => \$read2_file, -"log_file=s" => \$log_file, -"output1_file=s" => \$output1_file, -"output2_file=s" => \$output2_file, -"type=s" => \$TYPE, -"min_length=i" => \$MIN_LENGTH, -"min_quality=i" => \$MIN_QUALITY, -"verbose=s" => \$VERBOSE -) or die("Error in command line arguments\n"); - - -my $nb_read1=0; -my $nb_base_read1=0; -my $nb_read2=0; -my $nb_base_read2=0; - -my $nb_read1_t=0; -my $nb_base_read1_t=0; -my $nb_read2_t=0; -my $nb_base_read2_t=0; - -my $nb_base_current_t=0; - - -open(READ1, $read1_file) or die ("Can't open $read1_file\n"); -open(READ2, $read2_file) or die ("Can't open $read2_file\n"); -open(OUT1, ">$output1_file") or die ("Can't open $output1_file\n"); -open(OUT2, ">$output2_file") or die ("Can't open $output2_file\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 =<READ1>){ - my $ligne2_r1 =<READ1>; - my $ligne3_r1 =<READ1>; - my $ligne4_r1 =<READ1>; - my $ligne1_r2 =<READ2>; - my $ligne2_r2 =<READ2>; - my $ligne3_r2 =<READ2>; - my $ligne4_r2 =<READ2>; - - $nb_read1++; - $nb_read2++; - -#@ 1 sec - if ((!$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; - $nb_base_read1 += length($seq1); - } - if ($ligne2_r2 =~ /^([ATGCNX]+)\s*$/i){ - $seq2 = $1; - $nb_base_read2 += length($seq2); - } - 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=""; - my $nb_base_current_read1_t = 0; - my $nb_base_current_read2_t = 0; - - $fastq_lines_r1 = &grooming_and_trimming($ligne1_r1,$seq1,$qual1); - $nb_base_current_read1_t = $nb_base_current_t; - if ($fastq_lines_r1){ - $fastq_lines_r2 = &grooming_and_trimming($ligne1_r2,$seq2,$qual2); - $nb_base_current_read2_t = $nb_base_current_t; - } - if ($fastq_lines_r2){ - print OUT1 $fastq_lines_r1; - print OUT2 $fastq_lines_r2; - - $nb_read1_t++; - $nb_read2_t++; - $nb_base_read1_t += $nb_base_current_read1_t; - $nb_base_read2_t += $nb_base_current_read2_t; - - - } - } - } - - -#@ 7 sec - } -} - -close (READ1); -close (READ2); -close (OUT1); -close (OUT2); - -open (LF,">$log_file") or die("Can't open $log_file\n"); -print LF "\n####\t Fastq preparation \n"; -print LF "## Before preparation\n"; -print LF "#Read1 :\t$nb_read1\t#Base :\t$nb_base_read1\n"; -print LF "#Read2 :\t$nb_read2\t#Base :\t$nb_base_read2\n"; -print LF "## After preparation\n"; -print LF "#Read1 :\t$nb_read1_t\t#Base :\t$nb_base_read1_t\n"; -print LF "#Read2 :\t$nb_read2_t\t#Base :\t$nb_base_read2_t\n"; -close (LF); - - -sub grooming_and_trimming{ - my $header = shift; - my $seq = shift; - my $quality = shift; - my $quality_converted=""; - my $quality_ori=$quality; - - my $lengthseq = length($seq); - my $startTrim = 0; - my $stopTrim = length($quality)-1; - my $startnoN = $startTrim; - my $stopnoN = $stopTrim; - - - my $chercheN = $seq; - my @bad_position_N; - my @bad_position_Q; - my $current_index = index($chercheN,"N"); - my $abs_index = $current_index; - while ($current_index >=0){ - push (@bad_position_N,$abs_index); - - if ($current_index<length($seq)){ - $chercheN = substr($chercheN,$current_index+1); - $current_index = index($chercheN,"N"); - $abs_index = $current_index + $bad_position_N[$#bad_position_N]+1; - } - else { - last; - } - } - - my @q = split(//,$quality); - for (my $i=0;$i<=$#q;$i++){ - my $chr = $q[$i]; - my $num = ord($q[$i]); - if ($TYPE eq "illumina"){ - $num = $num - 31; # 31 comme la difference entre la plage sanger (33-> 93 / 0->60) et illumina (64->104 / 0->40) - $quality_converted .= chr($num); - } - - if ($num < $MIN_QUALITY + 33){ #33 comme le départ de la plage sanger - push(@bad_position_Q,$i); - } - } - if ($quality_converted){$quality = $quality_converted;} - - my @bad_position = (@bad_position_N, @bad_position_Q); - - if ($#bad_position>=0){ - @bad_position = sort {$a <=> $b} @bad_position; - my %coord=%{&extract_longer_string_coordinates_from_bad_position(0,$stopTrim,\@bad_position)}; - $startTrim = $coord{"start"}; - $stopTrim = $coord{"stop"}; -#print "$startTrim .. $stopTrim\n"; - - } - my $lengthTrim = $stopTrim - $startTrim +1; - - #if ($stats_length{$lengthTrim}){ - # $stats_length{$lengthTrim} = 1; - #} - #else { - # $stats_length{$lengthTrim}++; - #} - my $fastq_lines=""; - -# if ($header =~ /GA8\-EAS671_0005\:3\:1\:1043\:4432/){ -# print "HEAD:\t$header"; -# print "SEQ:\n$seq\n"; -# print "$quality_ori\n"; -# print "$quality\n"; -# for (my $i=0;$i<=$#bad_position;$i++){ -# print $bad_position[$i]."(".$q[$bad_position[$i]]." : ".ord($q[$bad_position[$i]]).")"."\t"; -# } -# print "\n"; -# print "$startTrim .. $stopTrim / $lengthTrim \n"; -# print $fastq_lines; -# print "\n"; -# } - - #for (my $i=$startTrim;$i<=$stopTrim;$i++){ - # if ($stats_quality{ord($q{$i])}){ - # $stats_quality{ord($q{$i])}=1; - # } - # else { - # $stats_quality{ord($q{$i])}++; - # } - #} - - if ($lengthTrim >= $MIN_LENGTH){ - $fastq_lines .= $header; - my $new_seq = substr($seq,$startTrim,$lengthTrim); -# $nb_base_current_t = length($new_seq); - $fastq_lines .= $new_seq."\n"; - $fastq_lines .= "+\n"; - my $new_q = substr($quality,$startTrim,$lengthTrim); - $fastq_lines .= $new_q."\n"; - return $fastq_lines; - - } - else { - #print "Insufficient length after trimming\n"; - return ""; - } -} - -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; -}
--- a/rapsodyn/PrepareFastqLight.xml Tue Oct 07 10:34:34 2014 -0400 +++ b/rapsodyn/PrepareFastqLight.xml Fri Oct 10 07:05:36 2014 -0400 @@ -1,4 +1,4 @@ -<tool id="PrepareFastqLight" name="PrepareFastqLight" version="1.01"> +<tool id="PrepareFastqLight" name="PrepareFastqLight" version="1.03"> <description>Fastq preparation</description> <command interpreter="perl"> PrepareFastqLight.pl -read1_file $input_read1_file -read2_file $input_read2_file -output1 $output_read1_file -output2 $output_read2_file -log_file $log_file -type $quality_type -min_quality $min_quality -min_length $min_length @@ -7,6 +7,7 @@ <param name="input_read1_file" type="data" format="txt,fastq" label="Select a suitable FASTQ READ 1 file from your history"/> <param name="input_read2_file" type="data" format="txt,fastq" label="Select a suitable FASTQ READ 2 file from your history"/> <param name="quality_type" type="select" label="Select input quality format"> + <option value="auto" selected="true">Auto-detect</option> <option value="sanger">Sanger</option> <option value="illumina">Illumina 1.3-1.7</option> </param>
--- a/rapsodyn/PrepareFastqLight.xml~ Tue Oct 07 10:34:34 2014 -0400 +++ /dev/null Thu Jan 01 00:00:00 1970 +0000 @@ -1,27 +0,0 @@ -<tool id="PrepareFastqLight" name="PrepareFastqLight" version="1.01"> -<description>Fastq preparation</description> -<command interpreter="perl"> - PrepareFastqLight.pl -read1_file $input_read1_file -read2_file $input_read2_file -output1 $output_read1_file -output2 $output_read2_file -log_file $log_file -type $quality_type -min_quality $min_quality -min_length $min_length -</command> -<inputs> - <param name="input_read1_file" type="data" format="txt,fastq" label="Select a suitable FASTQ READ 1 file from your history"/> - <param name="input_read2_file" type="data" format="txt,fastq" label="Select a suitable FASTQ READ 2 file from your history"/> - <param name="quality_type" type="select" label="Select input quality format"> - <option value="sanger">Sanger</option> - <option value="illumina">Illumina 1.3-1.7</option> - </param> - <param name="min_quality" type="integer" value="30" label="Minimum quality for 5' and 3' trimming "/> - <param name="min_length" type="integer" value="30" label="Minimum sequence length after trimming"/> -</inputs> -<outputs> - <data name="output_read1_file" format="fastqsanger" label="${tool.name} on ${on_string}"/> - <data name="output_read2_file" format="fastqsanger" label="${tool.name} on ${on_string}"/> - <data name="log_file" format="txt" label="${tool.name} LOG on ${on_string}"/> -</outputs> - -<help> - - - -</help> -</tool>
--- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/rapsodyn/listfiltering.pl Fri Oct 10 07:05:36 2014 -0400 @@ -0,0 +1,86 @@ +#!/usr/bin/perl +# v1.0.1 added log, and two different type of filtering (common / specific) +use strict; +use Getopt::Long; + +my $list1_file; +my $list2_file; +my $log_file; +my $NB_COL=1; +my $TYPE = "common"; +my %header; +my $nb_list1 = 0; +my $nb_list2 = 0; +my $nb_common = 0; + + +GetOptions ( +"list1_file=s" => \$list1_file, +"list2_file=s" => \$list2_file, +"log_file=s" => \$log_file, +"type=s" => \$TYPE, +"nb_col=i" => \$NB_COL +) or die("Error in command line arguments\n"); + +open(L2, $list2_file) or die("Can't open $list2_file\n"); +while (my $line=<L2>){ + $nb_list2++; + chomp($line); + my @fields = split(/\s+/,$line); + my $ref=""; + my $compt=0; + while ($compt<$NB_COL){ + if ($ref){$ref.="\t";} + $ref.=$fields[$compt]; + $compt++; + } + $header{$ref}=$line; +} +close (L2); + + +open(L1, $list1_file) or die("Can't open $list1_file\n"); +while (my $line=<L1>){ + $nb_list1++; + my @fields = split(/\s+/,$line); + my $ref=""; + my $compt=0; + while ($compt<$NB_COL){ + if ($ref){$ref.="\t";} + $ref.=$fields[$compt]; + $compt++; + } + # my $ref = "$fields[0]\t$fields[1]"; + + if ($header{$ref}){ + $nb_common++; + if ($TYPE eq "common"){ + print $line; + } + elsif ($TYPE eq "specific") { + } + else { + } + } + else { + if ($TYPE eq "common"){ + } + elsif ($TYPE eq "specific") { + print $line; + } + else { + } + } + +} +my $nb_list1_only = $nb_list1 - $nb_common; +my $nb_list2_only = $nb_list2 - $nb_common; + +close(L1); +open (LF,">$log_file") or die("Can't open $log_file\n"); +print LF "\n####\t List Filtering \n"; +print LF "#List 1 :\t$nb_list1 ($nb_list1_only)\n"; +print LF "#List 2 :\t$nb_list2 ($nb_list2_only)\n"; +print LF "#Common :\t$nb_common\n"; +close (LF); +
--- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/rapsodyn/listfiltering.xml Fri Oct 10 07:05:36 2014 -0400 @@ -0,0 +1,25 @@ +<tool id="listfiltering" name="listfiltering" version="1.01"> +<description>Compare list1 and list2</description> +<command interpreter="perl"> + listfiltering.pl -list1_file $input_list1 -list2_file $input_list2 -nb_col $nb_col -type $type -log_file $log_file > $output_file +</command> +<inputs> +<param name="input_list1" type="data" format="txt,pileup" label="Select list1 file from your history"/> +<param name="input_list2" type="data" format="txt,pileup" label="Select list2 file from your history"/> +<param name="type" type="select" label="Select type of filtering"> + <option value="specific">Only in list1</option> + <option value="common">Common</option> +</param> +<param name="nb_col" type="integer" value="2" label="Number of column to compare"/> +</inputs> +<outputs> + <data name="output_file" format="pileup" label="${tool.name} on ${on_string}"/> + <data name="log_file" format="txt" label="${tool.name} LOG on ${on_string}"/> +</outputs> + +<help> + + + +</help> +</tool>