diff megablast_xml_parser.py @ 0:35ff246876fc draft

Imported from capsule None
author devteam
date Mon, 19 May 2014 10:59:33 -0400
parents
children 3ce5d56297ed
line wrap: on
line diff
--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/megablast_xml_parser.py	Mon May 19 10:59:33 2014 -0400
@@ -0,0 +1,78 @@
+#!/usr/bin/env python
+    
+import sys, os, re
+
+if sys.version_info[:2] >= ( 2, 5 ):
+    import xml.etree.cElementTree as ElementTree
+else:
+    from galaxy import eggs
+    import pkg_resources; pkg_resources.require( "elementtree" )
+    from elementtree import ElementTree
+
+def stop_err( msg ):
+    sys.stderr.write( "%s\n" % msg )
+    sys.exit()
+
+def __main__():
+    source  = sys.argv[1]
+    hspTags = [
+           "Hsp_bit-score",
+           "Hsp_evalue",
+           "Hsp_query-from",
+           "Hsp_query-to",
+           "Hsp_hit-from",
+           "Hsp_hit-to",
+           "Hsp_query-frame",
+           "Hsp_hit-frame",
+           "Hsp_identity",
+           "Hsp_align-len",
+           "Hsp_qseq",
+           "Hsp_hseq",
+           "Hsp_midline"
+          ]
+    hspData = []
+
+    # get an iterable
+    try: 
+        context = ElementTree.iterparse( source, events=( "start", "end" ) )
+    except:
+        stop_err( "Invalid data format." )
+    # turn it into an iterator
+    context = iter( context )
+    # get the root element
+    try:
+        event, root = context.next()
+    except:
+        stop_err( "Invalid data format." )
+
+    outfile = open( sys.argv[2], 'w' )
+    try:
+        for event, elem in context:
+           # for every <Iteration> tag
+           if event == "end" and elem.tag == "Iteration":
+               query = elem.findtext( "Iteration_query-def" )
+               qLen = elem.findtext( "Iteration_query-len" )
+               # for every <Hit> within <Iteration>
+               for hit in elem.findall( "Iteration_hits/Hit" ):
+                   subject = hit.findtext( "Hit_id" )
+                   if re.search( '^gi', subject ):
+                       subject = subject.split('|')[1]
+                   sLen = hit.findtext( "Hit_len" )
+                   # for every <Hsp> within <Hit>
+                   for hsp in hit.findall( "Hit_hsps/Hsp" ):
+                        outfile.write( "%s\t%s\t%s\t%s" % ( query, qLen, subject, sLen ) )
+                        for tag in hspTags:
+                            outfile.write("\t%s" %(hsp.findtext( tag )))
+                            #hspData.append( hsp.findtext( tag ) )
+                        #hspData = []
+                        outfile.write('\n')
+               # prevents ElementTree from growing large datastructure
+               root.clear()
+               elem.clear()
+    except:
+        outfile.close()
+        stop_err( "The input data is malformed, or there is more than one dataset in the input file. Error: %s" % sys.exc_info()[1] )
+
+    outfile.close()
+
+if __name__ == "__main__": __main__()