| Previous changeset 3:3d18f8c3c0f6 (2020-06-05) Next changeset 5:75c279fa403a (2021-05-19) |
|
Commit message:
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ivar/ commit 693df287d23b0fd9dfd134b41d401a438c3f5ad6" |
|
modified:
ivar_removereads.xml |
|
added:
completemask.py sanitize_bed.py test-data/covid19/ARTIC-V1-bad.bed |
| b |
| diff -r 3d18f8c3c0f6 -r ee2beb764a7b completemask.py --- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/completemask.py Mon Jun 22 07:29:10 2020 -0400 |
| [ |
| @@ -0,0 +1,28 @@ +#!/usr/bin/env python + +import sys + + +if __name__ == '__main__': + with open(sys.argv[1]) as i: + getmasked_output = i.readline().strip() + + if not getmasked_output: + print() + print('No affected primer binding sites found!') + else: + masked_primers = getmasked_output.split('\t') + with open(sys.argv[2]) as i: + amplicon_data = [line.strip().split('\t') for line in i] + + masked_complete = [] + for primer in masked_primers: + for amplicon in amplicon_data: + if primer in amplicon: + masked_complete += amplicon + result = '\t'.join(sorted(set(masked_complete))) + print() + print('Removing reads primed with any of:') + print(result) + with open(sys.argv[1], 'w') as o: + o.write(result + '\n') |
| b |
| diff -r 3d18f8c3c0f6 -r ee2beb764a7b ivar_removereads.xml --- a/ivar_removereads.xml Fri Jun 05 04:11:42 2020 -0400 +++ b/ivar_removereads.xml Mon Jun 22 07:29:10 2020 -0400 |
| [ |
| @@ -1,24 +1,41 @@ -<tool id="ivar_removereads" name="ivar removereads" version="@VERSION@+galaxy0"> +<tool id="ivar_removereads" name="ivar removereads" version="@VERSION@+galaxy1"> <description>Remove reads from trimmed BAM file</description> <macros> <import>macros.xml</import> </macros> - <expand macro="requirements" /> + <expand macro="requirements"> + <requirement type="package" version="3.8.1">python</requirement> + </expand> <expand macro="version_command" /> <command detect_errors="exit_code"><![CDATA[ + cp '$input_bed' binding_sites.bed && + python '$__tool_directory__/sanitize_bed.py' binding_sites.bed && + + ivar getmasked + -i '$variants_tsv' -b binding_sites.bed -f '$amplicon_info' -p masked_primers && + + python '$__tool_directory__/completemask.py' masked_primers.txt '$amplicon_info' && ln -s '$input_bam' sorted.bam && - ln -s '$primer_index' primers.txt && - ln -s '$input_bed' bed.bed && + ln -s '${input_bam.metadata.bam_index}' sorted.bam.bai && + ivar removereads - -i sorted.bam + -i sorted.bam + -b binding_sites.bed -p removed_reads.bam - -t primers.txt - -b bed.bed - ]]> </command> + -t masked_primers.txt + ]]></command> <inputs> - <param name="input_bam" argument="-i" type="data" format="bam" label="Bam file" help="Aligned reads, to trim primers and quality"/> - <param name="primer_index" argument="-t" type="data" format="txt" label="Text file with primer indices separated by spaces" help="This is the output of getmasked command"/> - <param name="input_bed" argument="-b" type="data" format="bed" label="BED file with primer sequences and positions"/> + <param name="input_bam" argument="-i" type="data" format="bam" + label="Bam input" + help="BAM dataset, preprocessed with ivar trim, to remove reads from" /> + <param name="variants_tsv" type="data" format="tabular" + label="Variants input" + help="This dataset will be scanned for variants that affect primer binding sites and needs to be in tabular format with affected chromosome names in the first, and positions in the second column. If there is a header line, the name of the second column should be POS." /> + <param name="input_bed" argument="-b" type="data" format="bed" label="Primer binding sites information" + help="The same six-column BED dataset that served as input to ivar trim"/> + <param name="amplicon_info" type="data" format="tabular" + label="Primer to amplicon assignment info" + help="This input should consist of one line per amplicon with the tab-separated names of all primers used to generate that amplicon."/> </inputs> <outputs> <data name="output_bam" format="bam" label="${tool.name} on ${on_string}" from_work_dir="removed_reads.bam"/> @@ -26,20 +43,32 @@ <tests> <test> <param name="input_bam" value="zika/Z52_a.trimmed.sorted.bam"/> - <param name="primer_index" value="zika/primer_mismatchers_indices.txt"/> - <param name="input_bed" value="zika/db/zika_primers.bed"/> + <param name="variants_tsv" value="zika/primers_Z52_consensus.tsv"/> + <param name="input_bed" value="zika/db/zika_primers_consensus.bed"/> + <param name="amplicon_info" value="zika/db/pair_information.tsv"/> <output name="output_bam" file="zika/Z52_a.masked.bam" compare="sim_size" delta="100000" /> </test> </tests> <help><![CDATA[ - This command accepts an aligned and sorted BAM file trimmed using ivar trim - and removes the reads corresponding to the supplied primer indices, which is - the output of ivar getmasked command. Under the hood, ivar trim adds the - zero based primer index (based on the BED file) to the BAM auxillary data for - every read. Hence, ivar removereads will only work on BAM files that have - been trimmed using ivar trim. +This Galaxy tool combines the functionality of ``ivar getmasked`` and +``ivar removereads``. No separate ``ivar getmasked`` step is required when +using this tool. + +The wrapper takes as input a BAM dataset of aligned and sorted reads, from +which the primers listed in the primer binding sites BED input have been +trimmed with ``ivar trim``. + +From this input it will remove reads that come from amplicons that have been +generated with one or more primers that may have been affected in their binding +by variants listed in the variants input file. + +.. class:: Warning mark + + Preprocessing of the BAM input with ivar trim is essential for this tool to + work because only ``ivar trim`` can add required primer information to the + BAM auxillary data of every read. - Documentation can be found at `<https://andersen-lab.github.io/ivar/html/manualpage.html>`_. - ]]> </help> +ivar documentation can be found at `<https://andersen-lab.github.io/ivar/html/manualpage.html>`__. + ]]></help> <expand macro="citations" /> </tool> |
| b |
| diff -r 3d18f8c3c0f6 -r ee2beb764a7b sanitize_bed.py --- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/sanitize_bed.py Mon Jun 22 07:29:10 2020 -0400 |
| [ |
| @@ -0,0 +1,20 @@ +#!/usr/bin/env python + +import sys + + +with open(sys.argv[1]) as i: + bed_data = i.readlines() + +sanitized_data = [] +try: + for record in bed_data: + fields = record.split('\t') + sanitized_data.append( + '\t'.join(fields[:4] + ['60'] + fields[5:]) + ) +except IndexError: + pass # leave column number issue to getmasked +else: + with open(sys.argv[1], 'w') as o: + o.writelines(sanitized_data) |
| b |
| diff -r 3d18f8c3c0f6 -r ee2beb764a7b test-data/covid19/ARTIC-V1-bad.bed --- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/test-data/covid19/ARTIC-V1-bad.bed Mon Jun 22 07:29:10 2020 -0400 |
| b |
| b'@@ -0,0 +1,196 @@\n+MN908947.3\t30\t54\tnCoV-2019_1_LEFT\tNOINT\t+\n+MN908947.3\t385\t410\tnCoV-2019_1_RIGHT\tNOINT\t-\n+MN908947.3\t320\t342\tnCoV-2019_2_LEFT\tNOINT\t+\n+MN908947.3\t704\t726\tnCoV-2019_2_RIGHT\tNOINT\t-\n+MN908947.3\t642\t664\tnCoV-2019_3_LEFT\tNOINT\t+\n+MN908947.3\t1004\t1028\tnCoV-2019_3_RIGHT\tNOINT\t-\n+MN908947.3\t943\t965\tnCoV-2019_4_LEFT\tNOINT\t+\n+MN908947.3\t1312\t1337\tnCoV-2019_4_RIGHT\tNOINT\t-\n+MN908947.3\t1242\t1264\tnCoV-2019_5_LEFT\tNOINT\t+\n+MN908947.3\t1623\t1651\tnCoV-2019_5_RIGHT\tNOINT\t-\n+MN908947.3\t1573\t1595\tnCoV-2019_6_LEFT\tNOINT\t+\n+MN908947.3\t1942\t1964\tnCoV-2019_6_RIGHT\tNOINT\t-\n+MN908947.3\t1875\t1897\tnCoV-2019_7_LEFT\tNOINT\t+\n+MN908947.3\t2247\t2269\tnCoV-2019_7_RIGHT\tNOINT\t-\n+MN908947.3\t2181\t2205\tnCoV-2019_8_LEFT\tNOINT\t+\n+MN908947.3\t2568\t2592\tnCoV-2019_8_RIGHT\tNOINT\t-\n+MN908947.3\t2505\t2529\tnCoV-2019_9_LEFT\tNOINT\t+\n+MN908947.3\t2882\t2904\tnCoV-2019_9_RIGHT\tNOINT\t-\n+MN908947.3\t2826\t2850\tnCoV-2019_10_LEFT\tNOINT\t+\n+MN908947.3\t3183\t3210\tnCoV-2019_10_RIGHT\tNOINT\t-\n+MN908947.3\t3144\t3166\tnCoV-2019_11_LEFT\tNOINT\t+\n+MN908947.3\t3507\t3531\tnCoV-2019_11_RIGHT\tNOINT\t-\n+MN908947.3\t3460\t3482\tnCoV-2019_12_LEFT\tNOINT\t+\n+MN908947.3\t3826\t3853\tnCoV-2019_12_RIGHT\tNOINT\t-\n+MN908947.3\t3771\t3795\tnCoV-2019_13_LEFT\tNOINT\t+\n+MN908947.3\t4142\t4164\tnCoV-2019_13_RIGHT\tNOINT\t-\n+MN908947.3\t4054\t4077\tnCoV-2019_14_LEFT\tNOINT\t+\n+MN908947.3\t4428\t4450\tnCoV-2019_14_RIGHT\tNOINT\t-\n+MN908947.3\t4294\t4321\tnCoV-2019_15_LEFT\tNOINT\t+\n+MN908947.3\t4674\t4696\tnCoV-2019_15_RIGHT\tNOINT\t-\n+MN908947.3\t4636\t4658\tnCoV-2019_16_LEFT\tNOINT\t+\n+MN908947.3\t4995\t5017\tnCoV-2019_16_RIGHT\tNOINT\t-\n+MN908947.3\t4939\t4966\tnCoV-2019_17_LEFT\tNOINT\t+\n+MN908947.3\t5296\t5321\tnCoV-2019_17_RIGHT\tNOINT\t-\n+MN908947.3\t5230\t5259\tnCoV-2019_18_LEFT\tNOINT\t+\n+MN908947.3\t5620\t5644\tnCoV-2019_18_RIGHT\tNOINT\t-\n+MN908947.3\t5563\t5586\tnCoV-2019_19_LEFT\tNOINT\t+\n+MN908947.3\t5932\t5957\tnCoV-2019_19_RIGHT\tNOINT\t-\n+MN908947.3\t5867\t5894\tnCoV-2019_20_LEFT\tNOINT\t+\n+MN908947.3\t6247\t6272\tnCoV-2019_20_RIGHT\tNOINT\t-\n+MN908947.3\t6167\t6196\tnCoV-2019_21_LEFT\tNOINT\t+\n+MN908947.3\t6528\t6550\tnCoV-2019_21_RIGHT\tNOINT\t-\n+MN908947.3\t6466\t6495\tnCoV-2019_22_LEFT\tNOINT\t+\n+MN908947.3\t6846\t6873\tnCoV-2019_22_RIGHT\tNOINT\t-\n+MN908947.3\t6718\t6745\tnCoV-2019_23_LEFT\tNOINT\t+\n+MN908947.3\t7092\t7117\tnCoV-2019_23_RIGHT\tNOINT\t-\n+MN908947.3\t7035\t7058\tnCoV-2019_24_LEFT\tNOINT\t+\n+MN908947.3\t7389\t7415\tnCoV-2019_24_RIGHT\tNOINT\t-\n+MN908947.3\t7305\t7332\tnCoV-2019_25_LEFT\tNOINT\t+\n+MN908947.3\t7671\t7694\tnCoV-2019_25_RIGHT\tNOINT\t-\n+MN908947.3\t7626\t7651\tnCoV-2019_26_LEFT\tNOINT\t+\n+MN908947.3\t7997\t8019\tnCoV-2019_26_RIGHT\tNOINT\t-\n+MN908947.3\t7943\t7968\tnCoV-2019_27_LEFT\tNOINT\t+\n+MN908947.3\t8319\t8341\tnCoV-2019_27_RIGHT\tNOINT\t-\n+MN908947.3\t8249\t8275\tnCoV-2019_28_LEFT\tNOINT\t+\n+MN908947.3\t8635\t8661\tnCoV-2019_28_RIGHT\tNOINT\t-\n+MN908947.3\t8595\t8619\tnCoV-2019_29_LEFT\tNOINT\t+\n+MN908947.3\t8954\t8983\tnCoV-2019_29_RIGHT\tNOINT\t-\n+MN908947.3\t8888\t8913\tnCoV-2019_30_LEFT\tNOINT\t+\n+MN908947.3\t9245\t9271\tnCoV-2019_30_RIGHT\tNOINT\t-\n+MN908947.3\t9204\t9226\tnCoV-2019_31_LEFT\tNOINT\t+\n+MN908947.3\t9557\t9585\tnCoV-2019_31_RIGHT\tNOINT\t-\n+MN908947.3\t9477\t9502\tnCoV-2019_32_LEFT\tNOINT\t+\n+MN908947.3\t9834\t9858\tnCoV-2019_32_RIGHT\tNOINT\t-\n+MN908947.3\t9784\t9806\tnCoV-2019_33_LEFT\tNOINT\t+\n+MN908947.3\t10146\t10171\tnCoV-2019_33_RIGHT\tNOINT\t-\n+MN908947.3\t10076\t10099\tnCoV-2019_34_LEFT\tNOINT\t+\n+MN908947.3\t10437\t10459\tnCoV-2019_34_RIGHT\tNOINT\t-\n+MN908947.3\t10362\t10384\tnCoV-2019_35_LEFT\tNOINT\t+\n+MN908947.3\t10737\t10763\tnCoV-2019_35_RIGHT\tNOINT\t-\n+MN908947.3\t10666\t10688\tnCoV-2019_36_LEFT\tNOINT\t+\n+MN908947.3\t11048\t11074\tnCoV-2019_36_RIGHT\tNOINT\t-\n+MN908947.3\t10999\t11022\tnCoV-2019_37_LEFT\tNOINT\t+\n+MN908947.3\t11372\t11394\tnCoV-2019_37_RIGHT\tNOINT\t-\n+MN908947.3\t11306\t11331\tnCoV-2019_38_LEFT\tNOINT\t+\n+MN908947.3\t11668\t11693\tnCoV-2019_38_RIGHT\tNOINT\t-\n+MN908947.3\t11555\t11584\tnCoV-2019_39_LEFT\tNOINT\t+\n+MN908947.3\t11927\t11949\tnCoV-2019_39_RIGHT\tNOINT\t-\n+MN908947.3\t11863\t11889\tnCoV-2019_40_LEFT\tNOINT\t+\n+MN908947.3\t12234\t12256\tnCoV-2019_40_RIGHT\tNOINT\t-\n+MN908947.3\t12110\t12133\tnCoV-2019_41_LEFT\tNOINT\t+\n+MN908947.3\t12465\t12490\tnCoV-2019_41_RIGHT\tNOINT\t-\n+M'..b'\tNOINT\t+\n+MN908947.3\t18036\t18062\tnCoV-2019_59_RIGHT\tNOINT\t-\n+MN908947.3\t17966\t17993\tnCoV-2019_60_LEFT\tNOINT\t+\n+MN908947.3\t18324\t18348\tnCoV-2019_60_RIGHT\tNOINT\t-\n+MN908947.3\t18253\t18275\tnCoV-2019_61_LEFT\tNOINT\t+\n+MN908947.3\t18650\t18672\tnCoV-2019_61_RIGHT\tNOINT\t-\n+MN908947.3\t18596\t18618\tnCoV-2019_62_LEFT\tNOINT\t+\n+MN908947.3\t18957\t18979\tnCoV-2019_62_RIGHT\tNOINT\t-\n+MN908947.3\t18896\t18918\tnCoV-2019_63_LEFT\tNOINT\t+\n+MN908947.3\t19275\t19297\tnCoV-2019_63_RIGHT\tNOINT\t-\n+MN908947.3\t19204\t19232\tnCoV-2019_64_LEFT\tNOINT\t+\n+MN908947.3\t19591\t19616\tnCoV-2019_64_RIGHT\tNOINT\t-\n+MN908947.3\t19548\t19570\tnCoV-2019_65_LEFT\tNOINT\t+\n+MN908947.3\t19911\t19939\tnCoV-2019_65_RIGHT\tNOINT\t-\n+MN908947.3\t19844\t19866\tnCoV-2019_66_LEFT\tNOINT\t+\n+MN908947.3\t20231\t20255\tnCoV-2019_66_RIGHT\tNOINT\t-\n+MN908947.3\t20172\t20200\tnCoV-2019_67_LEFT\tNOINT\t+\n+MN908947.3\t20542\t20572\tnCoV-2019_67_RIGHT\tNOINT\t-\n+MN908947.3\t20472\t20496\tnCoV-2019_68_LEFT\tNOINT\t+\n+MN908947.3\t20867\t20890\tnCoV-2019_68_RIGHT\tNOINT\t-\n+MN908947.3\t20786\t20813\tnCoV-2019_69_LEFT\tNOINT\t+\n+MN908947.3\t21146\t21169\tnCoV-2019_69_RIGHT\tNOINT\t-\n+MN908947.3\t21075\t21104\tnCoV-2019_70_LEFT\tNOINT\t+\n+MN908947.3\t21427\t21455\tnCoV-2019_70_RIGHT\tNOINT\t-\n+MN908947.3\t21357\t21386\tnCoV-2019_71_LEFT\tNOINT\t+\n+MN908947.3\t21716\t21743\tnCoV-2019_71_RIGHT\tNOINT\t-\n+MN908947.3\t21658\t21682\tnCoV-2019_72_LEFT\tNOINT\t+\n+MN908947.3\t22013\t22038\tnCoV-2019_72_RIGHT\tNOINT\t-\n+MN908947.3\t21961\t21990\tnCoV-2019_73_LEFT\tNOINT\t+\n+MN908947.3\t22324\t22346\tnCoV-2019_73_RIGHT\tNOINT\t-\n+MN908947.3\t22262\t22290\tnCoV-2019_74_LEFT\tNOINT\t+\n+MN908947.3\t22626\t22650\tnCoV-2019_74_RIGHT\tNOINT\t-\n+MN908947.3\t22516\t22542\tnCoV-2019_75_LEFT\tNOINT\t+\n+MN908947.3\t22877\t22903\tnCoV-2019_75_RIGHT\tNOINT\t-\n+MN908947.3\t22797\t22819\tnCoV-2019_76_LEFT\tNOINT\t+\n+MN908947.3\t23192\t23214\tnCoV-2019_76_RIGHT\tNOINT\t-\n+MN908947.3\t23122\t23144\tnCoV-2019_77_LEFT\tNOINT\t+\n+MN908947.3\t23500\t23522\tnCoV-2019_77_RIGHT\tNOINT\t-\n+MN908947.3\t23443\t23466\tnCoV-2019_78_LEFT\tNOINT\t+\n+MN908947.3\t23822\t23847\tnCoV-2019_78_RIGHT\tNOINT\t-\n+MN908947.3\t23789\t23812\tnCoV-2019_79_LEFT\tNOINT\t+\n+MN908947.3\t24145\t24169\tnCoV-2019_79_RIGHT\tNOINT\t-\n+MN908947.3\t24078\t24100\tnCoV-2019_80_LEFT\tNOINT\t+\n+MN908947.3\t24443\t24467\tnCoV-2019_80_RIGHT\tNOINT\t-\n+MN908947.3\t24391\t24416\tnCoV-2019_81_LEFT\tNOINT\t+\n+MN908947.3\t24765\t24789\tnCoV-2019_81_RIGHT\tNOINT\t-\n+MN908947.3\t24696\t24721\tnCoV-2019_82_LEFT\tNOINT\t+\n+MN908947.3\t25052\t25076\tnCoV-2019_82_RIGHT\tNOINT\t-\n+MN908947.3\t24978\t25003\tnCoV-2019_83_LEFT\tNOINT\t+\n+MN908947.3\t25347\t25369\tnCoV-2019_83_RIGHT\tNOINT\t-\n+MN908947.3\t25279\t25301\tnCoV-2019_84_LEFT\tNOINT\t+\n+MN908947.3\t25646\t25673\tnCoV-2019_84_RIGHT\tNOINT\t-\n+MN908947.3\t25601\t25623\tnCoV-2019_85_LEFT\tNOINT\t+\n+MN908947.3\t25969\t25994\tnCoV-2019_85_RIGHT\tNOINT\t-\n+MN908947.3\t25902\t25924\tnCoV-2019_86_LEFT\tNOINT\t+\n+MN908947.3\t26290\t26315\tnCoV-2019_86_RIGHT\tNOINT\t-\n+MN908947.3\t26197\t26219\tnCoV-2019_87_LEFT\tNOINT\t+\n+MN908947.3\t26566\t26590\tnCoV-2019_87_RIGHT\tNOINT\t-\n+MN908947.3\t26520\t26542\tnCoV-2019_88_LEFT\tNOINT\t+\n+MN908947.3\t26890\t26913\tnCoV-2019_88_RIGHT\tNOINT\t-\n+MN908947.3\t26835\t26857\tnCoV-2019_89_LEFT\tNOINT\t+\n+MN908947.3\t27202\t27227\tnCoV-2019_89_RIGHT\tNOINT\t-\n+MN908947.3\t27141\t27164\tnCoV-2019_90_LEFT\tNOINT\t+\n+MN908947.3\t27511\t27533\tnCoV-2019_90_RIGHT\tNOINT\t-\n+MN908947.3\t27446\t27471\tnCoV-2019_91_LEFT\tNOINT\t+\n+MN908947.3\t27825\t27854\tnCoV-2019_91_RIGHT\tNOINT\t-\n+MN908947.3\t27784\t27808\tnCoV-2019_92_LEFT\tNOINT\t+\n+MN908947.3\t28145\t28172\tnCoV-2019_92_RIGHT\tNOINT\t-\n+MN908947.3\t28081\t28104\tnCoV-2019_93_LEFT\tNOINT\t+\n+MN908947.3\t28442\t28464\tnCoV-2019_93_RIGHT\tNOINT\t-\n+MN908947.3\t28394\t28416\tnCoV-2019_94_LEFT\tNOINT\t+\n+MN908947.3\t28756\t28779\tnCoV-2019_94_RIGHT\tNOINT\t-\n+MN908947.3\t28677\t28699\tnCoV-2019_95_LEFT\tNOINT\t+\n+MN908947.3\t29041\t29063\tnCoV-2019_95_RIGHT\tNOINT\t-\n+MN908947.3\t28985\t29007\tnCoV-2019_96_LEFT\tNOINT\t+\n+MN908947.3\t29356\t29378\tnCoV-2019_96_RIGHT\tNOINT\t-\n+MN908947.3\t29288\t29316\tnCoV-2019_97_LEFT\tNOINT\t+\n+MN908947.3\t29665\t29693\tnCoV-2019_97_RIGHT\tNOINT\t-\n+MN908947.3\t29486\t29510\tnCoV-2019_98_LEFT\tNOINT\t+\n+MN908947.3\t29836\t29866\tnCoV-2019_98_RIGHT\tNOINT\t-\n' |