|
0
|
1 <tool id="peptide_to_gff" name="Peptide to GFF" version="1.0">
|
|
|
2 <description>Map peptide to reference genome</description>
|
|
|
3 <requirements>
|
|
|
4 <requirement type="package">kent_tools</requirement>
|
|
|
5 <requirement type="package" version="master">peptide_to_gff</requirement>
|
|
|
6 </requirements>
|
|
|
7 <command>peptide_to_gff
|
|
|
8 #if $peptide_src.inputFormat == 'protein_pilot' or $peptide_src.inputFormat == 'tabular':
|
|
|
9 --skip_lines $peptide_src.input_file.metadata.comment_lines
|
|
|
10 --input $peptide_src.input_file
|
|
|
11 --peptide_column $peptide_src.peptide_col
|
|
|
12 --accession_column $peptide_src.accession_col
|
|
|
13 #end if
|
|
|
14 #for $ref in $ref_mappings:
|
|
|
15 #if $ref.mapping.ref_selector =='cdna_gtf':
|
|
|
16 --cdna_fasta $ref.mapping.seqs
|
|
|
17 --cdna_gtf $ref.mapping.feature_ref
|
|
|
18 #elif $ref.mapping.ref_selector =='cdna_ref':
|
|
|
19 --cdna_fasta $ref.mapping.seqs
|
|
|
20 #if $ref.mapping.genome.ref_src == 'indexed':
|
|
|
21 --reference $ref.mapping.genome.reference.fields.path
|
|
|
22 #else :
|
|
|
23 --reference $ref.mapping.genome.reference
|
|
|
24 #end if
|
|
|
25 #elif $ref.mapping.ref_selector =='cds_gff':
|
|
|
26 --cds_fasta $ref.mapping.seqs
|
|
|
27 --cds_gff $ref.mapping.feature_ref
|
|
|
28 #elif $ref.mapping.ref_selector =='eej':
|
|
|
29 --cdna_gtf $ref.mapping.feature_ref
|
|
|
30 #end if
|
|
|
31 #end for
|
|
|
32 #if $show_unmapped:
|
|
|
33 --unmapped $unmapped
|
|
|
34 #end if
|
|
|
35 --output $output_gff
|
|
|
36 </command>
|
|
|
37 <inputs>
|
|
|
38 <conditional name="peptide_src" >
|
|
|
39 <param name="inputFormat" type="select" label="Peptide Source Format">
|
|
|
40 <option value="protein_pilot">Protein Pilot Tabular</option>
|
|
|
41 <option value="tabular">Generic Tabular (with peptide and accession columns)</option>
|
|
|
42 <!-- future formats -->
|
|
|
43 </param>
|
|
|
44 <when value="protein_pilot">
|
|
|
45 <param name="input_file" type="data" format="tabular" label="Source File"
|
|
|
46 help="A tabular file that contains a peptide in a column and a accession name in another column."/>
|
|
|
47 <param name="peptide_col" type="hidden" value="13" label="Peptide Column"/>
|
|
|
48 <param name="accession_col" type="hidden" value="7" label="Accession Identifier Column"/>
|
|
|
49 <!--
|
|
|
50 <param name="info_cols" type="data_column" data_ref="input_file" label="Columns to include in the GTF INFO" multiple="true" optional="true"/>
|
|
|
51 -->
|
|
|
52 </when>
|
|
|
53 <when value="tabular">
|
|
|
54 <param name="input_file" type="data" format="tabular" label="Source File"
|
|
|
55 help="A tabular file that contains a peptide in a column and a accession name in another column."/>
|
|
|
56 <param name="peptide_col" type="data_column" data_ref="input_file" label="Peptide Column"/>
|
|
|
57 <param name="accession_col" type="data_column" data_ref="input_file" label="Accession Identifier Column"/>
|
|
|
58 <!--
|
|
|
59 <param name="info_cols" type="data_column" data_ref="input_file" label="Columns to include in the GTF INFO" multiple="true" optional="true"/>
|
|
|
60 -->
|
|
|
61 </when>
|
|
|
62 </conditional>
|
|
|
63 <repeat name="ref_mappings" title="Sequence and Feature References for mapping the peptides" min="1" help="" >
|
|
|
64 <conditional name="mapping" >
|
|
|
65 <param name="ref_selector" type="select" label="Select Peptide Mapping File Formats">
|
|
|
66 <option value="cdna_gtf">cdna sequence with GTF feature reference (Ensembl)</option>
|
|
|
67 <option value="cdna_ref">cdna sequence with genome sequence reference (ECgene)</option>
|
|
|
68 <option value="cds_gff">protein sequence with GFF feature reference (Augustus)</option>
|
|
|
69 <!-- Eventually may want to include putative exon-exon junctions
|
|
|
70 <option value="eej">putative exon exon junctions</option>
|
|
|
71 -->
|
|
|
72 </param>
|
|
|
73 <when value="cdna_gtf">
|
|
|
74 <param name="seqs" type="data" format="fasta" label="CDNA Transcripts Fasta"
|
|
|
75 help="Example: ftp://ftp.ensembl.org/pub/release-71/fasta/homo_saPiens/cdna/Homo_sapiens.GRCh37.71.cdna.all.fa.gz"/>
|
|
|
76 <param name="feature_ref" type="data" format="gtf" label=" GTF feature file for the cdna transcipts"
|
|
|
77 help="Example: ftp://ftp.ensembl.org/pub/release-71/gtf/homo_sapiens/Homo_sapiens.GRCh37.71.gtf.gz"/>
|
|
|
78 </when>
|
|
|
79 <when value="cdna_ref">
|
|
|
80 <param name="seqs" type="data" format="fasta" label="CDNA Transcripts Fasta"
|
|
|
81 help="Example: http://genome.ewha.ac.kr/ECgene/download/hg18/hg18_b1_low_fasta.txt.gz"/>
|
|
|
82 <conditional name="genome">
|
|
|
83 <param name="ref_src" type="select" label="Reference Genome Source for mapping">
|
|
|
84 <option value="indexed">Use a built-in index</option>
|
|
|
85 <option value="history">Use one in your history </option>
|
|
|
86 </param>
|
|
|
87 <when value="indexed">
|
|
|
88 <param name="reference" type="select" label="Genome Reference">
|
|
|
89 <options from_data_table="all_fasta">
|
|
|
90 </options>
|
|
|
91 </param>
|
|
|
92 </when>
|
|
|
93 <when value="history">
|
|
|
94 <param name="reference" type="data" format="fasta" label="Reference Genome Fasta"/>
|
|
|
95 </when>
|
|
|
96 </conditional>
|
|
|
97 </when>
|
|
|
98 <when value="cds_gff">
|
|
|
99 <param name="seqs" type="data" format="fasta" label="Protein Sequence Fasta"
|
|
|
100 help="Example: http://gbi.agrsci.dk/pig/sscrofa10_2_annotation/ssc10.2.RNA.hints.augustus.gff.prot.faa.gz"/>
|
|
|
101 <param name="feature_ref" type="data" format="gff3,gtf" label="GFF Feature file for the Protein Sequences"
|
|
|
102 help="Example: http://gbi.agrsci.dk/pig/sscrofa10_2_annotation/ssc10.2.RNA.hints.augustus.gff.gz"/>
|
|
|
103 </when>
|
|
|
104 <!-- Eventually may want to include putative exon-exon junctions
|
|
|
105 <when value="eej">
|
|
|
106 <param name="feature_ref" type="data" format="gtf" label=" The Ensembl GTF feature file corresponding to the EEJ build"
|
|
|
107 help="Example: ftp://ftp.ensembl.org/pub/release-71/gtf/homo_sapiens/Homo_sapiens.GRCh37.71.gtf.gz"/>
|
|
|
108 </when>
|
|
|
109 -->
|
|
|
110 </conditional>
|
|
|
111 </repeat>
|
|
|
112 <param name="show_unmapped" type="boolean" truevalue="yes" falsevalue="no" checked="False" label="Show unmapped lines"/>
|
|
|
113 </inputs>
|
|
|
114 <outputs>
|
|
|
115 <data format="gff3" metadata_source="input_file" name="output_gff" />
|
|
|
116 <data format_source="input_file" metadata_source="input_file" name="unmapped">
|
|
|
117 <filter>show_unmapped == True</filter>
|
|
|
118 </data>
|
|
|
119 </outputs>
|
|
|
120 <stdio>
|
|
|
121 <exit_code range="1:" level="fatal" description="Bad input dataset" />
|
|
|
122 </stdio>
|
|
|
123 <tests>
|
|
|
124 <test>
|
|
|
125 <param name="inputFormat" value="protein_pilot"/>
|
|
|
126 <param name="input_file" value="ProtData.tsv" ftype="tabular"/>
|
|
|
127 <!--
|
|
|
128 <param name="info_cols" value=""/>
|
|
|
129 -->
|
|
|
130 <param name="ref_selector" value="cdna_gtf"/>
|
|
|
131 <param name="seqs" value="GRCh37.69.cdna.19.fa" ftype="fasta"/>
|
|
|
132 <param name="feature_ref" value="Homo_sapiens.GRCh37_19.71.gtf" ftype="gtf"/>
|
|
|
133 <param name="show_unmapped" value="False"/>
|
|
|
134 <output name="output_gff" file="ProtData.gff3" ftype="gff3"/>
|
|
|
135 </test>
|
|
|
136 <!-- These values work when entered into a history, but aren't handled correctly by the function test environment.
|
|
|
137 <test>
|
|
|
138 <param name="inputFormat" value="protein_pilot"/>
|
|
|
139 <param name="input_file" value="ProtData.tsv" ftype="tabular"/>
|
|
|
140 <param name="ref_selector" value="cdna_ref"/>
|
|
|
141 <param name="seqs" value="GRCh37.69.cdna.19.fa" ftype="fasta"/>
|
|
|
142 <param name="ref_src" value="history"/>
|
|
|
143 <param name="reference" value="Homo_sapiens.GRCh37.71.dna.chromosome.19.fa" ftype="fasta"/>
|
|
|
144 <param name="show_unmapped" value="False"/>
|
|
|
145 <output name="output_gff" file="ProtData.gff3" ftype="gff3"/>
|
|
|
146 </test>
|
|
|
147 -->
|
|
|
148
|
|
|
149 </tests>
|
|
|
150 <help>
|
|
|
151 **Peptide to GFF**
|
|
|
152
|
|
|
153 Maps peptide sequences that have a known transcript or protein-coding sequence to a reference genome.
|
|
|
154
|
|
|
155 Inputs:
|
|
|
156
|
|
|
157 - A tabular file that contains a peptide column and a sequence accession column.
|
|
|
158
|
|
|
159 - One or more fasta files of transcript or protein sequences that match the accession column entries
|
|
|
160
|
|
|
161 - Either GTF, GFF, or a genome reference fasta for mapping the accession sequences to the genome.
|
|
|
162
|
|
|
163 Output:
|
|
|
164
|
|
|
165 - A GFF3 file with that gives the mapping of the peptide to the reference genome
|
|
|
166
|
|
|
167 - Optionally, a dataset with the unmapped lines from the input file.
|
|
|
168
|
|
|
169 </help>
|
|
|
170 </tool>
|
|
|
171
|
