Mercurial > repos > galaxyp > dbbuilder

comparison dbbuilder.xml @ 3:643d9192e3f5 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/dbbuilder commit bc3d2ee3b9fe72b2c11a19ac506e91cfd9146622
author galaxyp
date Wed, 03 May 2017 08:32:17 -0400
parents 0127e0bdf937
children 1bbb621c24f6
comparison
equal deleted inserted replaced
2:0127e0bdf937 3:643d9192e3f5
1 <tool id="dbbuilder" name="Protein Database Downloader" version="0.2.1"> 1 <tool id="dbbuilder" name="Protein Database Downloader" version="0.3.0">
2 <description></description> 2 <description></description>
3 <requirements> 3 <requirements>
4 <requirement type="package">gnu-wget</requirement> 4 <requirement type="package">gnu-wget</requirement>
5 </requirements> 5 </requirements>
6 <stdio> 6 <stdio>
18 #set $type = "direct" 18 #set $type = "direct"
19 #elif $source.from == "cRAP" 19 #elif $source.from == "cRAP"
20 ##set $url = "ftp://ftp.thegpm.org/fasta/cRAP/crap.fasta" 20 ##set $url = "ftp://ftp.thegpm.org/fasta/cRAP/crap.fasta"
21 #set $url = "https://raw.githubusercontent.com/pravs3683/cRAP/master/cRAP_protein_database.fasta" 21 #set $url = "https://raw.githubusercontent.com/pravs3683/cRAP/master/cRAP_protein_database.fasta"
22 #set $type = "direct" 22 #set $type = "direct"
23 #elif $source.from == "HMP"
24 #set $url = 'http://downloads.hmpdacc.org/data/reference_genomes/body_sites/' + str($source.site) + '.pep.fsa'
25 #set $type = "direct"
26 #elif $source.from == "HOMD"
27 #set $url = 'ftp://ftp.homd.org/human_oral_microbial_genomic_sequences/current/' + str($source.annotation)
28 #if str($source.annotation).endswith('.tar.gz'):
29 #set $type = "tgz"
30 #elif str($source.annotation).endswith('.zip'):
31 #set $type = "zip"
32 #end if
33 #elif $source.from == 'EBI Metagenomics'
34 #set $url = 'https://www.ebi.ac.uk/metagenomics/projects/' + str($source.ebi_project) + '/samples/' + str($source.ebi_sample) + '/runs/' + str($source.ebi_run) + '/results/versions/' + str($source.ebi_version) + '/sequences/' + str($source.ebi_annotation) + '/chunks/1'
35 #set $type = "gzip"
23 #elif $source.from == "url" 36 #elif $source.from == "url"
24 #set $url = $source.url 37 #set $url = $source.url
25 #set $type = "direct" 38 #set $type = $source.archive_type
26 #end if 39 #end if
27 #if $type =="direct" 40 #if $type =="direct"
28 wget -nv '$url' -O '${output_database}' 41 wget -nv '$url' -O '${output_database}' --no-check-certificate
42 #elif $type =="zip"
43 wget -nv '$url' -O tmp.zip --no-check-certificate && zcat -c tmp.zip > '${output_database}'
44 #elif $type =="gzip"
45 wget -nv '$url' -O tmp.gz --no-check-certificate && (if `command -v gzcat > /dev/null`; then gzcat tmp.gz; else zcat tmp.gz ; fi) > '${output_database}'
46 #elif $type =="bzip2"
47 wget -nv '$url' -O tmp.bz2 --no-check-certificate && bzcat tmp.bz2 > '${output_database}'
48 #elif $type =="tgz"
49 wget -nv '$url' -O tmp.tar.gz && tar zxfO tmp.tar.gz > '${output_database}'
50 #elif $type =="tbz"
51 wget -nv '$url' -O tmp.tar.bz && tar jxfO tmp.tar.bz > '${output_database}'
29 #end if 52 #end if
30 ]]> 53 ]]>
31 </command> 54 </command>
32 <inputs> 55 <inputs>
33 <conditional name="source"> 56 <conditional name="source">
34 <param name="from" type="select" label="Download from" help="Select database source. cRAP acts as a database for common MS contaminants. UniProtKB is a corss species collection of functional protein databases"> 57 <param name="from" type="select" label="Download from" help="Select database source. cRAP acts as a database for common MS contaminants. UniProtKB is a cross species collection of functional protein databases">
35 <option value="uniprot">UniProtKB</option> 58 <option value="uniprot">UniProtKB</option>
36 <option value="cRAP">cRAP (contaminants)</option> 59 <option value="cRAP">cRAP (contaminants)</option>
60 <option value="HMP">Human Microbiome Project body sites</option>
61 <option value="EBI Metagenomics">EBI Metagenomics</option>
62 <option value="HOMD">Human Oral Microbiome Database (HOMD)</option>
37 <option value="url">Custom URL</option> 63 <option value="url">Custom URL</option>
38 </param> 64 </param>
39 <when value="uniprot"> 65 <when value="uniprot">
40 <param name="taxon" type="select" format="text" help="select species for protein database"> 66 <param name="taxon" type="select" format="text" help="select species for protein database">
41 <label>Taxonomy</label> 67 <label>Taxonomy</label>
65 </sanitizer> 91 </sanitizer>
66 </param> 92 </param>
67 <param name="include_isoform" type="boolean" truevalue="&amp;include=yes" falsevalue="" label="Include isoform data" help="several different forms of a given protein are incorporated into database" /> 93 <param name="include_isoform" type="boolean" truevalue="&amp;include=yes" falsevalue="" label="Include isoform data" help="several different forms of a given protein are incorporated into database" />
68 </when> 94 </when>
69 <when value="cRAP" /> 95 <when value="cRAP" />
96 <when value="HMP">
97 <param name="site" type="select" label="Proteome for body site">
98 <option value="Airways">HMP airways</option>
99 <option value="Blood">HMP Blood</option>
100 <option value="Gastrointestinal_tract">HMP Gastro-intestinal tract</option>
101 <option value="Oral">HMP Oral</option>
102 <option value="Skin">HMP Skin</option>
103 <option value="Urogenital_tract">HMP urogenital Tract</option>
104 </param>
105 </when>
106 <when value="EBI Metagenomics">
107 <param name="ebi_project" value="" type="text" label="EBI Metagenomics Project (e.g. DRP003095)">
108 <help><![CDATA[
109 Look up EBI Metagenomics at https://www.ebi.ac.uk/metagenomics/search
110 ]]></help>
111 </param>
112 <param name="ebi_sample" value="" type="text" label="EBI Metagenomics Sample (e.g. DRS029200)">
113 </param>
114 <param name="ebi_run" value="" type="text" label="EBI Metagenomics Run (e.g. DRR033743)">
115 </param>
116 <param name="ebi_version" value="" type="text" label="EBI Metagenomics Pipeline Version (e.g. 3.0)">
117 </param>
118 <param name="ebi_annotation" type="boolean" label="With Annotation" checked="true"
119 truevalue="PredictedCDSWithAnnotation" falsevalue="PredictedCDSWithioutAnnotation"/>
120
121 </when>
122 <when value="HOMD">
123 <param name="annotation" type="select" label="Human Oral Microbiome Proteome">
124 <option value="oral_microbiome_dynamic.aa.zip">HOMD with dynamic annotation</option>
125 <option value="oral_microbiome.aa.tar.gz">HOMD with static annotation</option>
126 </param>
127 </when>
70 <when value="url"> 128 <when value="url">
71 <param name="url" value="" type="text" label="URL (http, ftp)"> 129 <param name="url" value="" type="text" label="URL (http, ftp) of Fasta sequences">
72 <sanitizer> 130 <sanitizer>
73 <valid> 131 <valid>
74 <add value="%"/> 132 <add value="%"/>
75 </valid> 133 </valid>
76 </sanitizer> 134 </sanitizer>
77 </param> 135 </param>
136 <param name="archive_type" type="select" label="Fasta source compression type">
137 <option value="direct" selected="true">fasta file (uncompressed)</option>
138 <option value="gzip">fasta.gz (gzip compressed)</option>
139 <option value="bzip2">fasta.bz2 (bzip2 compressed)</option>
140 <option value="zip">fasta.zip or fasta.Z (Zip compressed)</option>
141 <option value="tgz">fasta.tgz or fasta.tar.gz (tar archive gzip compressed)</option>
142 <option value="tbz">fasta.tbz or fasta.tar.bz (tar archive bzip2 compressed)</option>
143 </param>
78 </when> 144 </when>
79 </conditional> 145 </conditional>
80 </inputs> 146 </inputs>
81 <outputs> 147 <outputs>
82 <data format="fasta" name="output_database" label="Protein Database" /> 148 <data format="fasta" name="output_database" label="Protein Database ${source.from}" />
83 </outputs> 149 </outputs>
84 <tests> 150 <tests>
85 <test> 151 <test>
86 <param name="from" value="cRAP" /> 152 <param name="from" value="cRAP" />
87 <output name="output_database"> 153 <output name="output_database">
97 163
98 Creates a FASTA file of specified protein sequences for comparison with experimental MS/MS data in search algorithm. 164 Creates a FASTA file of specified protein sequences for comparison with experimental MS/MS data in search algorithm.
99 165
100 **External Links** 166 **External Links**
101 167
102 _Galaxy-P 101 shows usage Protein Database Downloader tool in the creation of a workflow 168 - Galaxy-P_101_ shows usage Protein Database Downloader tool in the creation of a workflow
103 .. _Galaxy-P 101: http://msi-galaxy-p.readthedocs.org/en/latest/sections/galaxyp_101.html 169 - UniProtKB_ provides additional information about the UniProt Knowledgebase
104 _UniProtKB provides additional information about the UniProt Knowledgebase 170
171
172 .. _Galaxy-P_101: http://msi-galaxy-p.readthedocs.org/en/latest/sections/galaxyp_101.html
105 .. _UniProtKB: http://www.uniprot.org/help/uniprotkb 173 .. _UniProtKB: http://www.uniprot.org/help/uniprotkb
174
175
176 **Additional Protein Fasta URLs**
177
178 *HUMAN GUT METAPROTEOME:*
179
180 * 512MB gzip ftp://public.genomics.org.cn/BGI/gutmeta/UniSet/UniGene.pep.gz
181 * 61MB gzip http://www.bork.embl.de/~arumugam/Qin_et_al_2010/frequent_microbe_proteins.fasta.gz
182
183
184 *MOUSE GUT MICROBIOTA:*
185
186 * 417MB gzip ftp://climb.genomics.cn/pub/10.5524/100001_101000/100114/Genecatalog/184sample_2.6M.GeneSet.pep.gz
187 * See: http://gigadb.org/dataset/view/id/100114/token/mZlMYJIF04LshpgP
188
189
106 ]]> 190 ]]>
107 </help> 191 </help>
192 <citations>
193 <citation type="doi">10.1093/nar/gkw1099</citation>
194 <citation type="doi">10.1093/nar/gkv1195 </citation>
195 <citation type="doi">10.1093/database/baq013</citation>
196 <citation type="doi">10.1038/nature11209</citation>
197 <citation type="doi">10.1038/nature11234</citation>
198 </citations>
108 </tool> 199 </tool>
109 200