From 94d04a1e503121a98b403f882c18a4f0799267d7 Mon Sep 17 00:00:00 2001
From: Calvin Morrison <mutantturkey@gmail.com>
Date: Wed, 29 Jan 2014 11:53:30 -0500
Subject: add filtering based on consecutive mer lengths

---
 src/filter_max_consecutive_binding.py | 72 +++++++++++++++++++++++++++++++++++
 1 file changed, 72 insertions(+)
 create mode 100755 src/filter_max_consecutive_binding.py

(limited to 'src/filter_max_consecutive_binding.py')

diff --git a/src/filter_max_consecutive_binding.py b/src/filter_max_consecutive_binding.py
new file mode 100755
index 0000000..daebee4
--- /dev/null
+++ b/src/filter_max_consecutive_binding.py
@@ -0,0 +1,72 @@
+#!/usr/bin/env python
+import sys, os
+
+binding = { 'A': 'T', 'T': 'A',	'C': 'G', 'G': 'C',	'_': False }
+
+
+def max_consecutive_binding(mer1, mer2):
+	if len(mer2) > len(mer1):
+		mer1, mer2 = mer2, mer1
+	
+	# reverse mer2,
+	mer2 = mer2[::-1]
+	# pad mer one to avoid errors
+	mer1 = mer1.ljust(len(mer1) + len(mer1), "_")
+
+	max_bind = 0;
+	for offset in range(len(mer2)):
+		consecutive = 0
+		for x in range(len(mer2)):
+			if binding[mer1[offset+x]] == mer2[x]:
+					consecutive = consecutive + 1
+			else:
+				consecutive = 0
+
+			max_bind = max(consecutive,max_bind)
+
+	return max_bind
+
+def test():
+	# mer 1  				 mer 2					# correct ans
+	arr = [
+	("ATATAT",			 "TATATA", 					5),
+	("ACAGGGAT",		 "ATATATAT", 				2),
+	("CATATATAT", 	 "ATATATATATAT",	  8),
+	("ATATATATATAT", "ATATATATAT", 		 10),
+	("ATATAT",			 "TATAT",					  5),
+	("AACGATACCATG", "GGATCATACGTA", 		3),
+	("CGT",          "ACG",			 				3),
+	("ACG",          "CGT",			 				3),
+	("CACC", 				 "GGTGT", 					4),
+	("GGTGT", 			 "CACC", 					4),
+	]
+	
+	print 'pass\tmer1\tmer2\tres\tcorr'
+	for mer_combination in arr:
+		response = []
+		ans = max_consecutive_binding(mer_combination[0], mer_combination[1])
+	
+		response.append(str(ans == mer_combination[2]))
+		response.append(mer_combination[0])
+		response.append(mer_combination[1])
+		response.append(str(ans))
+		response.append(str(mer_combination[2]))
+	
+		print '\t'.join(response)
+	
+def main():
+
+	if(len(sys.argv) < 2):
+		print "cutoff is expected as an argument"
+		exit()
+	else:
+		cutoff = int(sys.argv[1])
+	
+	for line in sys.stdin:
+		mer = line.split()[0]
+		if max_consecutive_binding(mer, mer) < cutoff:
+			sys.stdout.write(line)
+
+
+if __name__ == "__main__":
+	sys.exit(main())
-- 
cgit v1.2.3