On branch temp

data.py, utils.py: get_chrom_format(chrom) -> get_chrom_format(chroms) instead of taking one chrom string, now takes a list of all the chroms
uci-cbcl · Jan 9, 2014 · 1dfa9af · 1dfa9af
1 parent be15d17
commit 1dfa9af
Show file tree

Hide file tree

Showing 2 changed files with 22 additions and 12 deletions.
diff --git a/pyloh/preprocess/data.py b/pyloh/preprocess/data.py
@@ -103,7 +103,7 @@ def segmentation_by_chrom(self, normal_bam, tumor_bam):
         chrom_num = len(chrom_ID_list)
 
         sam_SQ = normal_bam.header['SQ']
-        sam_chrom_format = get_chrom_format(sam_SQ[0]['SN'])
+        sam_chrom_format = get_chrom_format(map(lambda x:x['SN'], sam_SQ))
         chrom_lens = self._get_chrom_lens(chrom_ID_list, sam_SQ)
 
         for i in range(0, chrom_num):
@@ -128,11 +128,11 @@ def segmentation_by_bed(self, normal_bam, tumor_bam, bed_file_name):
         chrom_start = constants.CHROM_START
 
         sam_SQ = normal_bam.header['SQ']
-        sam_chrom_format = get_chrom_format(sam_SQ[0]['SN'])
+        sam_chrom_format = get_chrom_format(map(lambda x:x['SN'], sam_SQ))
         chrom_lens = self._get_chrom_lens(chrom_ID_list, sam_SQ)
 
         bed_chroms, bed_starts, bed_ends = BEDParser(bed_file_name)
-        bed_chrom_format = get_chrom_format(bed_chroms[0])
+        bed_chrom_format = get_chrom_format(bed_chroms)
         bed_num = len(bed_chroms)
 
         for i in range(0, bed_num):

diff --git a/pyloh/preprocess/utils.py b/pyloh/preprocess/utils.py
@@ -54,16 +54,26 @@ def chrom_name_to_ID(chrom_name):
 
     return ID
 
-def get_chrom_format(chrom):
-    if chrom[0:3] == 'chr':
-        return 'UCSC'
+def get_chrom_format(chroms):
+    format = 'NONE'
+
+    for chrom in chroms:
+        if chrom[0:3] == 'chr':
+            format = 'UCSC'
+            break
+        else:
+            try:
+                ID = int(chrom)
+                format = 'ENSEMBL'
+                break
+            except:
+                pass
+
+    if format == 'NONE':
+        print 'Error: %s not supported' % (chrom)
+        sys.exit(-1)
     else:
-        try:
-            ID = int(chrom)
-            return 'ENSEMBL'
-        except:
-            print 'Error: %s not supported' % (chrom)
-            sys.exit(-1)
+        return format
 
 def normal_heterozygous_filter(counts):
     BAF_N_MAX = constants.BAF_N_MAX