# HG changeset patch
# User xuebing
# Date 1333214744 14400
# Node ID ce94ba22a3b36abc480674f84ca6bf7bd2fdaf51
# Parent  082c1e8ff413a35fffd739451e7be92e17f7fb81
Uploaded

diff -r 082c1e8ff413 -r ce94ba22a3b3 bed_collapse.py
--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/bed_collapse.py	Sat Mar 31 13:25:44 2012 -0400
@@ -0,0 +1,36 @@
+'''
+collapse intervals
+'''
+
+def collapseInterval_strand(filename,c_strand,c_score):
+    # keeping max column c
+    uniqintv = {}
+    data = {}
+    f = open(filename)
+    header = f.readline()
+    if 'chr' in header:
+        flds = header.strip().split('\t')
+        key = '\t'.join([flds[0],flds[1],flds[2],flds[c_strand]])
+        uniqintv[key] = float(flds[c_score])
+        data[key] = flds
+    for line in f:
+        flds = line.strip().split('\t')
+        key = '\t'.join([flds[0],flds[1],flds[2],flds[c_strand]])
+        if not uniqintv.has_key(key):
+            uniqintv[key] = float(flds[c_score])
+            data[key] = flds
+        elif uniqintv[key] < float(flds[c_score]):
+            uniqintv[key] = float(flds[c_score])
+            data[key] = flds
+            
+    f.close()        
+    for key in uniqintv.keys():
+        print '\t'.join(data[key])
+        
+import sys
+
+if sys.argv[2] == '0':#ignore strand
+    sys.argv[2] = 1
+if sys.argv[3] == '0':# ignore score
+    sys.argv[3] = 2
+collapseInterval_strand(sys.argv[1],int(sys.argv[2])-1,int(sys.argv[3])-1)