Mercurial > repos > michal_klinka > trtools

comparison StatSTR/statSTR.xml @ 0:ecf8c4f9f8ba draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
"planemo upload for repository https://github.com/Kulivox/TRToolsGalaxyWrapper commit 08480e3145858e41b14bf1061f87243e8cc2d04e"
author michal_klinka
date Sat, 23 Apr 2022 16:34:03 +0000
parents
children 8e8cf43f6836
comparison
equal deleted inserted replaced
-1:000000000000 0:ecf8c4f9f8ba
1 <?xml version="1.0" ?>
2 <tool id="statSTR" name="StatSTR" version="@tool_version@">
3
4 <description>StatSTR takes in a TR genotyping VCF file and outputs per-locus statistics.</description>
5
6 <macros>
7 <import>macros.xml</import>
8 </macros>
9 <expand macro="requirements"/>
10 <version_command>statSTR --version</version_command>
11
12 <command detect_errors="exit_code"><![CDATA[statSTR
13
14 ## vcf definition
15 #if $input_output.vcf:
16 --vcf $input_output.vcf
17 #end if
18 ## end vcf definition
19 ## out definition
20 #if $input_output.out:
21 --out $input_output.out
22 #end if
23 ## end out definition
24 ## vcftype definition
25 #if $input_output.vcftype:
26 --vcftype $input_output.vcftype
27 #end if
28 ## end vcftype definition
29 ## precision definition
30 #if $input_output.precision:
31 --precision $input_output.precision
32 #end if
33 ## end precision definition
34 ## samples definition
35 #if $filtering_group.samples:
36 --samples $filtering_group.samples
37 #end if
38 ## end samples definition
39 ## sample_prefixes definition
40 #if $filtering_group.sample_prefixes:
41 --sample-prefixes $filtering_group.sample_prefixes
42 #end if
43 ## end sample_prefixes definition
44 ## region definition
45 #if $filtering_group.region:
46 --region $filtering_group.region
47 #end if
48 ## end region definition
49 ## thresh definition
50 #if $stats_group.thresh:
51 --thresh
52 #end if
53 ## end thresh definition
54 ## afreq definition
55 #if $stats_group.afreq:
56 --afreq
57 #end if
58 ## end afreq definition
59 ## acount definition
60 #if $stats_group.acount:
61 --acount
62 #end if
63 ## end acount definition
64 ## hwep definition
65 #if $stats_group.hwep:
66 --hwep
67 #end if
68 ## end hwep definition
69 ## het definition
70 #if $stats_group.het:
71 --het
72 #end if
73 ## end het definition
74 ## entropy definition
75 #if $stats_group.entropy:
76 --entropy
77 #end if
78 ## end entropy definition
79 ## mean definition
80 #if $stats_group.mean:
81 --mean
82 #end if
83 ## end mean definition
84 ## mode definition
85 #if $stats_group.mode:
86 --mode
87 #end if
88 ## end mode definition
89 ## var definition
90 #if $stats_group.var:
91 --var
92 #end if
93 ## end var definition
94 ## numcalled definition
95 #if $stats_group.numcalled:
96 --numcalled
97 #end if
98 ## end numcalled definition
99 ## use_length definition
100 #if $stats_group.use_length:
101 --use-length
102 #end if
103 ## end use_length definition
104 ## plot_afreq definition
105 #if $plotting_group.plot_afreq:
106 --plot-afreq
107 #end if
108 ## end plot_afreq definition
109 ## version definition
110 #if $version9ac8.version:
111 --version
112 #end if
113 ## end version definition
114 1>$stdout 2>$stderr
115 ]]></command>
116
117
118 <inputs>
119 <section name="input_output" title="Input/output" expanded="false">
120 <param argument="--vcf" type="data" format="vcf" optional="false" label="vcf" help="Input STR VCF file"/>
121 <param argument="--out" type="text" optional="false" label="out"
122 help="Output file prefix. Use stdout to print file to standard output. In addition, if not stdout then timing diagnostics are print to stdout."/>
123 <param argument="--vcftype" type="select" optional="true" label="vcftype"
124 help="Genotyper that was used to create this VCF input">
125 <expand macro="vcfTypes"/>
126 </param>
127 <param argument="--precision" type="integer" optional="true" label="precision"
128 help="How much precision to use when printing decimals"/>
129 </section>
130 <section name="filtering_group" title="Filtering group" expanded="false">
131 <param argument="--samples" type="text" optional="true" label="samples"
132 help="File containing list of samples to include. Or a comma-separated list of files to compute stats separate for each group of samples"/>
133 <param argument="--sample-prefixes" type="text" optional="true" label="sample_prefixes"
134 help="Prefixes to name output for each samples group. By default uses 1,2,3 etc."/>
135 <param argument="--region" type="text" optional="true" label="region"
136 help="Restrict to the region chrom:start-end. Requires file to bgzipped and tabix indexed."/>
137 </section>
138 <section name="stats_group" title="Stats group" expanded="false">
139 <param argument="--thresh" type="boolean" optional="true" label="thresh"
140 help="Output threshold field (max allele size, used for GangSTR strinfo)."/>
141 <param argument="--afreq" type="boolean" optional="true" label="afreq" help="Output allele frequencies"/>
142 <param argument="--acount" type="boolean" optional="true" label="acount" help="Output allele counts"/>
143 <param argument="--hwep" type="boolean" optional="true" label="hwep" help="Output HWE p-values per loci."/>
144 <param argument="--het" type="boolean" optional="true" label="het"
145 help="Output the heterozygosity of each locus."/>
146 <param argument="--entropy" type="boolean" optional="true" label="entropy"
147 help="Output the entropy of each locus."/>
148 <param argument="--mean" type="boolean" optional="true" label="mean"
149 help="Output mean of the allele frequencies."/>
150 <param argument="--mode" type="boolean" optional="true" label="mode"
151 help="Output mode of the allele frequencies."/>
152 <param argument="--var" type="boolean" optional="true" label="var"
153 help="Output variance of the allele frequencies."/>
154 <param argument="--numcalled" type="boolean" optional="true" label="numcalled"
155 help="Output number of samples called."/>
156 <param argument="--use-length" type="boolean" optional="true" label="use_length"
157 help="Calculate per-locus stats (het, HWE) collapsing alleles by length. This is implicitly true for genotypers which only emit length based genotypes."/>
158 </section>
159 <section name="plotting_group" title="Plotting group" expanded="false">
160 <param argument="--plot-afreq" type="boolean" optional="true" label="plot_afreq"
161 help="Output allele frequency plot. Will only do for a maximum of 10 TRs."/>
162 </section>
163 <section name="version9ac8" title="Version" expanded="false">
164 <param argument="--version" type="boolean" optional="true" label="version"
165 help="show program's version number and exit"/>
166 </section>
167 </inputs>
168
169
170 <outputs>
171 <data name="stdout" label="Outputs and console output" format="txt">
172 <discover_datasets format="tabular" pattern="(?P&lt;designation&gt;.+)\.tab" visible="true"/>
173 </data>
174 <data name="stderr" label="stderr output" format="txt"/>
175 </outputs>
176
177
178 <tests>
179 <test>
180 <section name="main_params">
181 <param name="vcf" value="NA12878_chr21_hipstr.sorted.vcf.gz"/>
182 <param name="out" value="test_result"/>
183 <param name="vcftype" value="hipstr"/>
184 </section>
185 <section name="opt_basic">
186 <param name="precision" value="5"/>
187 </section>
188 <section name="statistic_opt">
189 <param name="thresh" value="true"/>
190 <param name="afreq" value="true"/>
191 <param name="acount" value="true"/>
192 <param name="hwep" value="true"/>
193 <param name="het" value="true"/>
194 <param name="entropy" value="true"/>
195 <param name="mean" value="true"/>
196 <param name="mode" value="true"/>
197 <param name="var" value="true"/>
198 <param name="numcalled" value="true"/>
199 </section>
200
201 <output name="stdout">
202 <assert_contents>
203 <has_text text="Finished 9550 records"/>
204 </assert_contents>
205 <discovered_dataset designation="test_result" ftype="tabular" file="test_result.tab"/>
206 </output>
207
208 <output name="stderr" file="stderr.txt"/>
209
210 </test>
211 </tests>
212
213
214 <help>usage: main.py [-h] --vcf VCF --out OUT [--vcftype VCFTYPE] [--precision PRECISION] [--samples SAMPLES]
215 [--sample-prefixes SAMPLE_PREFIXES] [--region REGION] [--thresh] [--afreq] [--acount] [--hwep] [--het]
216 [--entropy] [--mean]
217 [--mode] [--var] [--numcalled] [--use-length] [--plot-afreq] [--version]
218
219 options:
220 -h, --help show this help message and exit
221
222 Input/output:
223 --vcf VCF Input STR VCF file
224 --out OUT Output file prefix. Use stdout to print file to standard output. In addition, if not stdout then
225 timing diagnostics are print to stdout.
226 --vcftype VCFTYPE ##!!## Name trh could not be loaded
227 --precision PRECISION
228 How much precision to use when printing decimals
229
230 Filtering group:
231 --samples SAMPLES File containing list of samples to include. Or a comma-separated list of files to compute
232 stats separate for each group of samples
233 --sample-prefixes SAMPLE_PREFIXES
234 Prefixes to name output for each samples group. By default uses 1,2,3 etc.
235 --region REGION Restrict to the region chrom:start-end. Requires file to bgzipped and tabix indexed.
236
237 Stats group:
238 --thresh Output threshold field (max allele size, used for GangSTR strinfo).
239 --afreq Output allele frequencies
240 --acount Output allele counts
241 --hwep Output HWE p-values per loci.
242 --het Output the heterozygosity of each locus.
243 --entropy Output the entropy of each locus.
244 --mean Output mean of the allele frequencies.
245 --mode Output mode of the allele frequencies.
246 --var Output variance of the allele frequencies.
247 --numcalled Output number of samples called.
248 --use-length Calculate per-locus stats (het, HWE) collapsing alleles by length. This is implicitly true for
249 genotypers which only emit length based genotypes.
250
251 Plotting group:
252 --plot-afreq Output allele frequency plot. Will only do for a maximum of 10 TRs.
253
254 Version:
255 --version show program's version number and exit
256 </help>
257
258
259 <citations>
260 <citation type="bibtex">@misc{TRTools: a toolkit for genome-wide analysis of tandem repeats,
261 author = {Nima Mousavi, Jonathan Margoliash, Neha Pusarla, Shubham Saini, Richard Yanicky, Melissa Gymrek},
262 year = {2020},
263 title = {TRTools},
264 publisher = {GitHub},
265 journal = {GitHub repository},
266 url = {https://github.com/gymreklab/trtools},
267 }
268 </citation>
269 </citations>
270 </tool>