changeset 1:757f1fc216f7 default tip

Uploaded

--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/bed_resize.py	Sat Mar 31 13:19:44 2012 -0400
@@ -0,0 +1,57 @@
+'''
+change start and end of interval files
+'''
+
+import sys
+
+def resize(infile,outfile,expr_start,expr_end,strand):
+    fin = open(infile)
+    fout = open(outfile,'w')
+    if expr_start[0:3] == 'end':
+        c1 = 2
+        n1 = int(expr_start[3:])
+    else:
+        c1 = 1
+        n1 = int(expr_start[5:])
+    if expr_end[0:3] == 'end':
+        c2 = 2
+        n2 = int(expr_end[3:])
+    else:
+        c2 = 1
+        n2 = int(expr_end[5:])
+    if strand == 'ignore':
+        for line in fin:
+            flds = line.strip().split('\t')
+            start = int(flds[c1]) + n1
+            if start >= 0:
+                end = int(flds[c2]) + n2
+                if end >= start:
+                    flds[1] = str(start)
+                    flds[2] = str(end)
+                    fout.write('\t'.join(flds)+'\n')
+    else:# upstream downstream
+       for line in fin:
+            flds = line.strip().split('\t')
+            if flds[5] == '+':
+                start = int(flds[c1]) + n1
+                if start >= 0:
+                    end = int(flds[c2]) + n2
+                    if end >= start: 
+                        flds[1] = str(start)
+                        flds[2] = str(end)
+                        fout.write('\t'.join(flds)+'\n')
+            else: # on the - strand
+                start = int(flds[3-c2]) - n2 # end=end+n2
+                if start >= 0:
+                    end = int(flds[3-c1]) - n1
+                    if end >= start:
+                        flds[1] = str(start)
+                        flds[2] = str(end)
+                        fout.write('\t'.join(flds)+'\n')
+    fin.close()
+    fout.close()
+
+if __name__ == "__main__":
+    resize(sys.argv[1],sys.argv[2],sys.argv[3],sys.argv[4],sys.argv[5])
+    # python resize.py in.bed out.bed start-3 end+5 both
+    # python resize.py <input.bed> <output.bed> expr_start expr_end strand(both/+/-)