From fca7e7de2846f2c7a99d41b96d554813a1890e7f Mon Sep 17 00:00:00 2001
From: Greg Pinero <gregpinero@gmail.com>
Date: Fri, 5 Jul 2013 11:37:40 -0400
Subject: [PATCH] Update grepfqparser.py

Fixed grep to remove -- lines and removed extra code.
---
 grepfqparser.py | 52 +++++++++++++++++++++----------------------------
 1 file changed, 22 insertions(+), 30 deletions(-)

diff --git a/grepfqparser.py b/grepfqparser.py
index 1922790..34e6908 100644
--- a/grepfqparser.py
+++ b/grepfqparser.py
@@ -43,21 +43,24 @@
 def main():
         #parse command line options
         try:
-                opts, arg = getopt.getopt(sys.argv[1:],"h", ["help"])
+                opts, arg = getopt.getopt(sys.argv[1:],"ht:", ["help"])
         except getopt.error, msg:
                 print msg
                 print "for help use --help"
                 sys.exit(2)
         # process options
+        offset = 0
         for o, a in opts:
+                if o == '-t':
+                    offset = int(a)
+                    print "using offset", offset
                 if o in ("-h", "--help"):
                         print __doc__
                         sys.exit(0)
         if len(arg) < 3:
-                print "\nUsage: python grepfqparser.py <input_fastq> <barcode_file> <output_folder>\m"                
+                print "\nUsage: python grepfqparser.py [options] <input_fastq> <barcode_file> <output_folder>\m"                
                 sys.exit(0)
         #process arguments
-
         fqFile = arg[0]
         bcFile = arg[1]
         OutFolder = arg[2]
@@ -83,7 +86,7 @@ def main():
         else:
                 gzbool  = "NO"
                 
-        print "fq file gzipped? = %s" %(gzbool)        
+        print "fq file gzipped? = %s" %(gzbool)
         if gzbool == "YES":
                 print "unzipping file"
                 tempfq = open("tempfq",'w')
@@ -101,31 +104,18 @@ def main():
                 barcode_up = barcode.upper()
                 name = lineItems[1]
                 print barcode
-                
-                parsed_file_step1_name = str(OutFolder + "/indiv" + name + "_" + barcode + "firstgrep")
-                parsed_file_step1 = open(parsed_file_step1_name,'w') 
-                errlog = open("errlog2",'w')
-                cmd = 'grep -B 1 -A 2 ^%s %s' % (barcode_up, fqFile)
-                subprocess.call(cmd, shell=True,stdout=parsed_file_step1, stderr=errlog)
-                #(note: pipe into sed to remove barcodes and associated quality scores from each line)
-                cmd = "grep -B 1 -A 2 ^%s %s | sed '2~2s/^%s//g'" % (barcode_up, fqFile, '.'*len(barcode_up))
-                subprocess.call(cmd, shell=True,stdout=parsed_file_step1, stderr=errlog)
-                errlog.close()
-                parsed_file_step1.close()
-                
-                """grep with -B and -A produces spacer marks '--' in file. Cannot figure out how to suppress these, 
-                so remove and paste into new file, then delete original"""
-                
                 parsed_file_name = str(OutFolder + "/indiv" + name + "_" + barcode)
                 parsed_file = open(parsed_file_name,'w')
-                errlog = open("errlog3",'w')
-                cmd = 'awk "!/^--$/" %s' % (parsed_file_step1_name)
-                subprocess.check_call(cmd,shell=True,stdout=parsed_file,stderr=errlog)
-                errlog.close()
-                parsed_file.close()
-                cmd = 'rm %s' % (parsed_file_step1_name)
-                subprocess.check_call(cmd,shell=True)
-        
+                errlog = open("errlog2",'w')
+                #First grep finds lines starting with the barcode and includes the line above, and two lines below each match
+                #Pipe into grep again to filter out -- between matches which some (versions??) of grep insert
+                #Pipe into sed to remove barcodes (optional offset) and associated quality scores from each line
+                cmd = """grep -B 1 -A 2 ^%s %s | grep -v "^--$" | sed '2~2s/^%s//g'""" % (barcode_up, fqFile, '.'*(len(barcode_up)+offset))
+                try:
+                    failed = subprocess.call(cmd, shell=True,stdout=parsed_file, stderr=errlog)
+                finally:
+                    errlog.close()
+                    parsed_file.close()
         bc.close()
   
         """Now collect all unparsed reads"""
@@ -146,9 +136,11 @@ def main():
         nomatch_file = open(OutFolder + "/nomatches",'w')
         errlog = open("errlog4",'w')
         cmd = "awk 'NR%%4==2' %s | grep -f %s -v" % (fqFile, bconly)
-        nomatch = subprocess.call(cmd, shell=True, stdout=nomatch_file,stderr=errlog)
-        errlog.close()
-        nomatch_file.close()
+        try:
+            nomatch = subprocess.call(cmd, shell=True, stdout=nomatch_file,stderr=errlog)
+        finally:
+            errlog.close()
+            nomatch_file.close()
         
         """delete tempfq, the gunzipped original file"""
         cmd = 'rm tempfq bcOnly errlog1 errlog2 errlog3 errlog4'