comparison ensembl_variant_report.xml @ 0:9f4ea174ce3d draft

planemo upload for repository https://github.com/jj-umn/galaxytools/tree/master/ensembl_variant_report commit e6aa05bbbee3cc7d98f16354fc41c674f439ff1b-dirty
author jjohnson
date Thu, 14 Jun 2018 17:51:39 -0400
parents
children f87fe6bc48f4
comparison
equal deleted inserted replaced
-1:000000000000 0:9f4ea174ce3d
1 <tool id="ensembl_variant_report" name="Ensembl Variant Report" version="0.1.0">
2 <requirements>
3 <requirement type="package" version="1.40">gtf_to_genes</requirement>
4 <requirement type="package" version="3.1.4">twobitreader</requirement>
5 <requirement type="package" version="1.62">biopython</requirement>
6 </requirements>
7 <stdio>
8 <exit_code range="1:" />
9 </stdio>
10 <command><![CDATA[
11 python '$__tool_directory__/ensembl_variant_report.py'
12 #if $ref.ref_source == 'cached':
13 --twobit="$ref.ref_loc.fields.path"
14 #else
15 --twobit="$ref.ref_file"
16 #end if
17 --gene_model="$gtf_file"
18 #if $variant.fmt == 'vcf':
19 --input="$variant.input_vcf"
20 --format=snpeff
21 #else
22 --input="$variant.input_tsv"
23 --pos_column=$variant.pos_column
24 --ref_column=$variant.ref_column
25 --alt_column=$variant.alt_column
26 --transcript_column=$variant.transcript_column
27 --dp_column=$variant.dp_column
28 --dpr_column=$variant.dpr_column
29 #end if
30 #if str($filter.min_depth) != '':
31 --min_depth=$filter.min_depth
32 #end if
33 #if str($filter.min_freq) != '':
34 --min_freq=$filter.min_freq
35 #end if
36 #if str($report.readthrough) != '':
37 --readthrough=$report.readthrough
38 #end if
39 #if str($report.leading_aa) != '':
40 --leading_aa=$report.leading_aa
41 #end if
42 #if str($report.trailing_aa) != '':
43 --trailing_aa=$report.trailing_aa
44 #end if
45 --output="$output"
46 ]]></command>
47 <inputs>
48 <conditional name="variant">
49 <param name="fmt" type="select" label="Input format for variants">
50 <option value="vcf">snpEff vcf</option>
51 <option value="tsv">tabular from snpsift extract</option>
52 </param>
53 <when value="vcf">
54 <param name="input_vcf" type="data" format="vcf" label="snpEff VCF file"/>
55 </when>
56 <when value="tsv">
57 <param name="input_tsv" type="data" format="tabular" label="tabular file"/>
58 <param name="pos_column" type="data_column" data_ref="input_tsv" label="POS column"/>
59 <param name="ref_column" type="data_column" data_ref="input_tsv" label="REF column"/>
60 <param name="alt_column" type="data_column" data_ref="input_tsv" label="ALT column"/>
61 <param name="transcript_column" type="data_column" data_ref="input_tsv" label="Transcript ID column"/>
62 <param name="dp_column" type="data_column" data_ref="input_tsv" label="Read Depth (DP) column"/>
63 <param name="dpr_column" type="data_column" data_ref="input_tsv" label="Allele Count (DPR ro AN) column"/>
64 </when>
65 </conditional>
66 <conditional name="ref">
67 <param name="ref_source" type="select" label="Source for Genomic Data">
68 <option value="cached">Locally cached</option>
69 <option value="history">History</option>
70 </param>
71 <when value="cached">
72 <param name="ref_loc" type="select" label="Select reference 2bit file">
73 <options from_data_table="twobit" />
74 </param>
75 </when>
76 <when value="history">
77 <param name="ref_file" type="data" format="twobit" label="reference 2bit file" />
78 </when>
79 </conditional>
80 <param name="gtf_file" type="data" format="gtf" label="Ensembl GTF file"/>
81 <section name="filter" expanded="false" title="Filter Options">
82 <param name="min_depth" type="integer" value="" optional="true" min="0" label="Minimum Read Depth to report"
83 help="If a value is entered, ignore variants with DP INFO field less than this value"/>
84 <param name="min_freq" type="float" value="" optional="true" min="0.0" max="1.0" label="Minimum Alt frequency to report"
85 help="If a value is entered, ignore variant alleles where its DPR count divided the sum of DPR is less than this value"/>
86 </section>
87 <section name="report" expanded="false" title="Report Options">
88 <param name="readthrough" type="integer" value="0" optional="true" min="0" max="4" label="Number of readthrough stop codons to display"/>
89 <param name="leading_aa" type="integer" value="10" optional="true" min="0" label="Number of Amino Acids prior to variant to display"
90 help="Ignored for frame shifts"/>
91 <param name="trailing_aa" type="integer" value="10" optional="true" min="0" label="Number of Amino Acids following the variant to display"
92 help="Ignored for frame shifts"/>
93 </section>
94
95 </inputs>
96 <outputs>
97 <data name="output" format="tabular" />
98 </outputs>
99 <tests>
100 <test>
101 <param name="fmt" value="vcf"/>
102 <param name="input_vcf" value="GRCh38_X400k_ANN.vcf" ftype="vcf"/>
103 <param name="ref_source" value="history"/>
104 <param name="ref_file" value="GRCh38_X.2bit" ftype="twobit"/>
105 <param name="gtf_file" value="GRCh38_X.83.gtf" ftype="gtf"/>
106 <output name="output">
107 <assert_contents>
108 <has_text text="MGGQ_A_SASNSFSRLH" />
109 <not_has_text text="*REDHAAGPEA" />
110 </assert_contents>
111 </output>
112 </test>
113 <test>
114 <param name="fmt" value="vcf"/>
115 <param name="input_vcf" value="GRCh38_X400k_EFF.vcf" ftype="vcf"/>
116 <param name="ref_source" value="history"/>
117 <param name="ref_file" value="GRCh38_X.2bit" ftype="twobit"/>
118 <param name="gtf_file" value="GRCh38_X.83.gtf" ftype="gtf"/>
119 <output name="output">
120 <assert_contents>
121 <has_text text="MGGQ_A_SASNSFSRLH" />
122 <not_has_text text="*REDHAAGPEA" />
123 </assert_contents>
124 </output>
125 </test>
126 <test>
127 <param name="fmt" value="vcf"/>
128 <param name="input_vcf" value="GRCh38_X400k_ANN.vcf" ftype="vcf"/>
129 <param name="ref_source" value="history"/>
130 <param name="ref_file" value="GRCh38_X.2bit" ftype="twobit"/>
131 <param name="gtf_file" value="GRCh38_X.83.gtf" ftype="gtf"/>
132 <param name="readthrough" value="1"/>
133 <output name="output">
134 <assert_contents>
135 <has_text text="MGGQ_A_SASNSFSRLH" />
136 <has_text text="*REDHAAGPEA" />
137 </assert_contents>
138 </output>
139 </test>
140 <test>
141 <param name="fmt" value="vcf"/>
142 <param name="input_vcf" value="GRCh38_X400k_ANN.vcf" ftype="vcf"/>
143 <param name="ref_source" value="history"/>
144 <param name="ref_file" value="GRCh38_X.2bit" ftype="twobit"/>
145 <param name="gtf_file" value="GRCh38_X.83.gtf" ftype="gtf"/>
146 <param name="min_depth" value="100"/>
147 <param name="min_freq" value=".80"/>
148 <output name="output">
149 <assert_contents>
150 <has_text text="EQLDAGVRYL_E_LRIAHMLEGS" />
151 <not_has_text text="EYDILVAEET_V_GEPWEDGFEA" />
152 </assert_contents>
153 </output>
154 </test>
155 <test>
156 <param name="fmt" value="tsv"/>
157 <param name="input_tsv" value="GRCh38_X400k.tsv" ftype="tabular"/>
158 <param name="ref_source" value="history"/>
159 <param name="ref_file" value="GRCh38_X.2bit" ftype="twobit"/>
160 <param name="gtf_file" value="GRCh38_X.83.gtf" ftype="gtf"/>
161 <param name="pos_column" value="2"/>
162 <param name="ref_column" value="3"/>
163 <param name="alt_column" value="4"/>
164 <param name="transcript_column" value="7"/>
165 <param name="dp_column" value="8"/>
166 <param name="dpr_column" value="9"/>
167 <output name="output">
168 <assert_contents>
169 <has_text text="MGGQ_A_SASNSFSRLH" />
170 <not_has_text text="*REDHAAGPEA" />
171 </assert_contents>
172 </output>
173 </test>
174 </tests>
175 <help><![CDATA[
176 Uses an Essembl GTF and a genome 2bit reference to report variant peptides from snpEff reported missense and frameshift variants.
177 Allows readthrough of stop codons, and reports the stop codons.
178
179 Input can be a snpEff vcf file using either ANN or EFF annotations.
180 Alternatively, the input can be a tabular file that has columns:
181
182 - pos
183 - ref
184 - alt
185 - Ensembl Transcript ID
186 - Read Depth (DP)
187 - AlleleDepth (DPR)
188
189 ]]></help>
190 <citations>
191 <citation type="doi">10.1093/bioinformatics/btt385</citation>
192 </citations>
193 </tool>