|
0
|
1 <tool id="vcftools_subset" name="Subset" version="0.1">
|
|
|
2 <description>columns from a VCF dataset</description>
|
|
|
3
|
|
|
4 <requirements>
|
|
|
5 <requirement type="package">tabix</requirement>
|
|
|
6 <requirement type="package" version="0.1.11">vcftools</requirement>
|
|
|
7 </requirements>
|
|
|
8
|
|
|
9 <command>
|
|
|
10 ln -s ${input} input.vcf ;
|
|
|
11
|
|
|
12 ## Sort file.
|
|
|
13 sort -k1,1 -k2,2n input.vcf > input.vcf.sorted ;
|
|
|
14
|
|
|
15 ## Compress.
|
|
|
16 bgzip input.vcf.sorted ;
|
|
|
17
|
|
|
18 ## Index.
|
|
|
19 tabix -p vcf input.vcf.sorted.gz ;
|
|
|
20
|
|
|
21 ## Subset.
|
|
|
22 vcf-subset ${trim_alt_alleles} ${exclude_rows_with_no_variants} -c "${columns}" input.vcf.sorted.gz > $output
|
|
|
23 </command>
|
|
|
24
|
|
|
25 <inputs>
|
|
|
26 <param name="input" label="Input dataset" type="data" format="vcf"/>
|
|
|
27 <param name="columns" label="Columns" type="text" value=""/>
|
|
|
28 <param name="trim_alt_alleles" type="select" label="Remove alternate alleles if not found in the subset" help="">
|
|
|
29 <option value="" selected="True">No</option>
|
|
|
30 <option value="-a">Yes</option>
|
|
|
31 </param>
|
|
|
32 <param name="exclude_rows_with_no_variants" type="select" label="Exclude rows not containing variants" help="">
|
|
|
33 <option value="" selected="True">No</option>
|
|
|
34 <option value="-e">Yes</option>
|
|
|
35 </param>
|
|
|
36 </inputs>
|
|
|
37
|
|
|
38 <outputs>
|
|
|
39 <data name="output" format="vcf"/>
|
|
|
40 </outputs>
|
|
|
41
|
|
|
42 <stdio>
|
|
|
43 <regex match=".*" source="both" level="log" description="tool progress"/>
|
|
|
44 </stdio>
|
|
|
45
|
|
|
46 <tests>
|
|
|
47 <test>
|
|
|
48 <param name="input" value="subset_in.vcf"/>
|
|
|
49 <param name="columns" value="sample1"/>
|
|
|
50 <param name="trim_alt_alleles" value="Yes"/>
|
|
|
51 <param name="exclude_rows_with_no_variants" value="Yes"/>
|
|
|
52 <!-- 2 lines diff because command line with full file path is included in output VCF, and
|
|
|
53 it not possible to match full file path. -->
|
|
|
54 <output name="output" file="subset_out.vcf" lines_diff="2"/>
|
|
|
55 </test>
|
|
|
56 </tests>
|
|
|
57
|
|
|
58 <help>
|
|
|
59 Please see the VCFtools `documentation`__ for help and further information.
|
|
|
60
|
|
|
61 .. __: http://vcftools.sourceforge.net/docs.html
|
|
|
62 </help>
|
|
|
63 </tool>
|
