diff tools/dbbuilder.xml @ 4:b39347891609 draft

Uploaded

--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/tools/dbbuilder.xml	Fri Sep 26 14:31:06 2014 -0400
@@ -0,0 +1,95 @@
+<tool id="dbbuilder" version="0.2.0" name="Protein Database Downloader">
+  <description>
+  </description>
+  <requirements>
+    <requirement type="binary">wget</requirement>    
+  </requirements>  
+  <stdio>
+    <exit_code range="1:"  level="fatal" description="Error downloading database." />
+    <regex match="ERROR" level="fatal" source="stderr" description="Error downloading database." />
+  </stdio>
+  <!-- TODO: escape quotes. -->
+    <!-- Add NCBI and maxquant contaminants. -->
+    <!-- http://maxquant.org/contaminants.zip -->
+    <!-- ftp://ftp.ncbi.nih.gov/refseq/H_sapiens/mRNA_Prot/human.protein.faa.gz-->
+  <command>
+    #if $source.from == "uniprot"
+    #set $url = "http://www.uniprot.org/uniprot/?query=taxonomy%3a%22" + str($source.taxon) + "%22" + str($source.set) + str($source.reviewed) + "&amp;force=yes&amp;format=fasta" + str($source.include_isoform)
+    #set $type = "direct"
+    #elif $source.from == "cRAP"
+    #set $url = "ftp://ftp.thegpm.org/fasta/cRAP/crap.fasta"
+    #set $type = "direct"
+    #elif $source.from == "url"
+    #set $url = $source.url
+    #set $type = "direct"
+    #end if
+    #if $type =="direct"
+    wget -nv '$url' -O '${output_database}'
+    #end if
+  </command>
+  <inputs>
+    <conditional name="source">
+  <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">
+        <option value="uniprot">UniProtKB</option>
+        <option value="cRAP">cRAP (contaminants)</option>
+        <option value="url">Custom URL</option>
+      </param>
+      <when value="uniprot">
+        <param name="taxon" type="select" format="text" help="select species for protein database">
+          <label>Taxonomy</label>
+          <options from_file="uniprot_taxons.loc">
+            <column name="name" index="0" />
+            <column name="value" index="1" />
+          </options>
+        </param>
+        <param name="reviewed" type="select" help="UniProtKB/TrEMBL (unreviewed)is a large, automatically annotated database- may contain redundant sequences, but there is a higher chance peptides will be identified. UniProtKB/Swiss-Prot (reviewed) is a smaller, manually annotated database- less of a chance peptides will be identified but less sequence redundancy">
+          <option value="+">UniProtKB</option>
+          <option value="+reviewed%3Ayes">UniProtKB/Swiss-Prot (reviewed only)</option>
+          <option value="+reviewed%3Ano">UniProtKB/TrEMBL (unreviewed only)</option>
+          <sanitizer>
+            <valid>
+              <add value="%"/>
+            </valid>
+          </sanitizer>
+        </param>
+        <param name="set" type="select" label="Proteome Set">
+          <option value="+">Any</option>
+          <option value="+keyword%3a1185" selected="true">Reference Proteome Set</option>
+          <option value="+keyword%3a181">Complete Proteome Set</option>
+          <sanitizer>
+            <valid>
+              <add value="%"/>
+            </valid>
+          </sanitizer>
+        </param>
+        <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" />
+      </when>
+      <when value="cRAP" />
+      <when value="url">
+        <param name="url" value="" type="text" label="URL (http, ftp)">
+          <sanitizer>
+            <valid>
+              <add value="%"/>
+            </valid>
+          </sanitizer>
+        </param>
+      </when>
+    </conditional>
+  </inputs>
+  <outputs>
+    <data format="fasta" name="output_database" label="Protein Database" />
+  </outputs>
+  <help>
+**Output**
+
+Creates a FASTA file of specified protein sequences for comparison with experimental MS/MS data in search algorithm. 
+
+**External Links**
+
+_Galaxy-P 101 shows usage Protein Database Downloader tool in the creation of a workflow
+.. _Galaxy-P 101: http://msi-galaxy-p.readthedocs.org/en/latest/sections/galaxyp_101.html
+_UniProtKB provides additional information about the UniProt Knowledgebase
+.. _UniProtKB: http://www.uniprot.org/help/uniprotkb
+  </help>
+</tool>
+