changeset 2:6c624803c5f9 draft

Update.

--- a/dbbuilder.xml	Wed Dec 19 00:07:18 2012 -0500
+++ b/dbbuilder.xml	Fri May 10 17:09:36 2013 -0400
@@ -1,25 +1,38 @@
-<tool id="dbbuilder" version="0.1.0" name="Protein Database Downloader">
+<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 computing FDR" />
+    <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+keyword%3a1185&amp;force=yes&amp;format=fasta" + str($source.include_isoform)
+    #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">
         <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" >
@@ -29,8 +42,38 @@
             <column name="value" index="1" />
           </options>
         </param>
+        <param name="reviewed" type="select">
+          <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" />
       </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>

--- a/uniprot_taxons.loc.sample	Wed Dec 19 00:07:18 2012 -0500
+++ b/uniprot_taxons.loc.sample	Fri May 10 17:09:36 2013 -0400
@@ -5,3 +5,6 @@
 Sus scrofa (Pig)	9823
 Escherichia coli (strain K12)	83333
 Arabidopsis thaliana (Mouse-ear cress)	3702
+Caenorhabditis elegans (C.elegans)	6239
+Drosophila melanogaster (Fruit fly)	7227
+Saccharomyces cerevisiae (Baker's Yeast)	4932
\ No newline at end of file