diff find_in_reference.py @ 0:fe0327a3ba81

Uploaded
author jjohnson
date Sat, 04 Jan 2014 09:03:57 -0500
parents
children 856033fb26e8
line wrap: on
line diff
--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/find_in_reference.py	Sat Jan 04 09:03:57 2014 -0500
@@ -0,0 +1,123 @@
+#!/usr/bin/env python
+"""
+#
+#------------------------------------------------------------------------------
+#                         University of Minnesota
+#         Copyright 2013, Regents of the University of Minnesota
+#------------------------------------------------------------------------------
+# Author:
+#
+#  James E Johnson
+#
+#------------------------------------------------------------------------------
+"""
+
+"""
+Takes 2 tabular files as input:  
+  1. The file to be filtered 
+  2. The reference file 
+
+The string value of selected column of the input file is searched for 
+in the string values of the selected column of the reference file.
+
+The intended purpose is to filter a peptide fasta file in tabular format 
+by whether those peptide sequences are found in a reference fasta file.
+
+"""
+import sys,re,os.path
+import tempfile
+import optparse
+from optparse import OptionParser
+import logging
+
+
+def __main__():
+  #Parse Command Line
+  parser = optparse.OptionParser()
+  parser.add_option( '-i', '--input', dest='input', help='The input file to filter. (Otherwise read from stdin)' )
+  parser.add_option( '-r', '--reference', dest='reference', help='The reference file to filter against' )
+  parser.add_option( '-o', '--output', dest='output', help='The output file for input lines filtered by reference')
+  parser.add_option( '-f', '--filtered', dest='filtered', help='The output file for input lines not in the output')
+  parser.add_option('-c','--input_column', dest='input_column', default=None, help='The column for the value in the input file. (first column = 1, default to last column)')
+  parser.add_option('-C','--reference_column', dest='reference_column', default=None, help='The column for the value in the reference file. (first column = 1, default to last column)')
+  parser.add_option( '-k', '--keep', dest='keep', action="store_true", default=False, help='' )
+  parser.add_option( '-d', '--debug', dest='debug', action='store_true', default=False, help='Turn on wrapper debugging to stdout'  )
+  (options, args) = parser.parse_args()
+  # Input files
+  if options.input != None:
+    try:
+      inputPath = os.path.abspath(options.input)
+      inputFile = open(inputPath, 'r')
+    except Exception, e:
+      print >> sys.stderr, "failed: %s" % e
+      exit(2)
+  else:
+    inputFile = sys.stdin
+  # Reference
+  if options.reference == None:
+      print >> sys.stderr, "failed: reference file is required" 
+      exit(2)
+  # Output files
+  outFile = None
+  filteredFile = None
+  if options.filtered == None and options.output == None:
+    #write to stdout
+    outFile = sys.stdout
+  else:
+    if options.output != None:
+      try:
+        outPath = os.path.abspath(options.output)
+        outFile = open(outPath, 'w')
+      except Exception, e:
+        print >> sys.stderr, "failed: %s" % e
+        exit(3)
+    if options.filtered != None:
+      try:
+        filteredPath = os.path.abspath(options.filtered)
+        filteredFile = open(filteredPath, 'w')
+      except Exception, e:
+        print >> sys.stderr, "failed: %s" % e
+        exit(3)
+  incol = -1
+  if options.input_column and options.input_column > 0:
+    incol = int(options.input_column)-1
+  refcol = -1
+  if options.reference_column and options.reference_column > 0:
+    refcol = int(options.reference_column)-1
+  refFile = None
+  for ln,line in enumerate(inputFile):
+    try:
+      found = False
+      search_string = line.split('\t')[incol].rstrip('\r\n')
+      if options.debug: 
+        print >> sys.stderr, "search: %s" % (search_string)
+      refFile = open(options.reference,'r')
+      for tn,fline in enumerate(refFile):
+        target_string = fline.split('\t')[refcol]
+        if options.debug: 
+          print >> sys.stderr, "in: %s %s %s" % (search_string,search_string in target_string,target_string)
+        if search_string in target_string:
+          found = True
+          break
+      if found:
+        if options.keep == True:
+          if outFile:
+              outFile.write(line)
+        else:
+          if filteredFile:
+            filteredFile.write(line)
+      else:
+        if options.keep == True:
+          if filteredFile:
+            filteredFile.write(line)
+        else:
+          if outFile:
+            outFile.write(line)
+    except Exception, e:
+      print >> sys.stderr, "failed: Error reading %s - %s" % (options.reference,e)
+    finally:
+      if refFile:
+        refFile.close()
+
+if __name__ == "__main__" : __main__()
+