fix a bug that reduce recall rate

tjiangHIT · Nov 15, 2019 · 17225ed · 17225ed
1 parent 112f855
commit 17225ed
Showing 1 changed file with 27 additions and 27 deletions.
diff --git a/src/cuteSV/cuteSV b/src/cuteSV/cuteSV
@@ -165,33 +165,33 @@ def analysis_split_read(split_read, SV_size, RLength, read_name, candidate):
 
 
 
-	# if len(SP_list) >= 3 and trigger_INS_TRA == 1:
-	# 	if SP_list[0][4] == SP_list[-1][4]:
-	# 		# print(SP_list[0])
-	# 		# print(SP_list[-1])
-	# 		if SP_list[0][5] != SP_list[-1][5]:
-	# 			pass
-	# 		else:
-	# 			if SP_list[0][5] == '+':
-	# 				ele_1 = SP_list[0]
-	# 				ele_2 = SP_list[-1]
-	# 			else:
-	# 				ele_1 = [RLength-SP_list[-1][1], RLength-SP_list[-1][0]]+SP_list[-1][2:]
-	# 				ele_2 = [RLength-SP_list[0][1],RLength-SP_list[0][0]]+SP_list[0][2:]
-	# 			# print(ele_1)
-	# 			# print(ele_2)
-	# 			dis_ref = ele_2[2] - ele_1[3]
-	# 			dis_read = ele_2[0] - ele_1[1]
-	# 			if dis_ref < 100 and dis_read - dis_ref >= SV_size and dis_read - dis_ref <= 100000:
-	# 				# print(min(ele_2[2], ele_1[3]), dis_read - dis_ref, read_name)
-	# 				if ele_1[4] not in candidate['INS']:
-	# 					candidate['INS'][ele_1[4]] = list()
-	# 				candidate["INS"][ele_2[4]].append([min(ele_2[2], ele_1[3]), dis_read - dis_ref, read_name])
-
-	# 			if dis_ref <= -SV_size:
-	# 				if ele_2[4] not in candidate["DUP"]:
-	# 					candidate["DUP"][ele_2[4]] = list()
-	# 				candidate["DUP"][ele_2[4]].append([ele_2[2], ele_1[3], read_name])
+	if len(SP_list) >= 3 and trigger_INS_TRA == 1:
+		if SP_list[0][4] == SP_list[-1][4]:
+			# print(SP_list[0])
+			# print(SP_list[-1])
+			if SP_list[0][5] != SP_list[-1][5]:
+				pass
+			else:
+				if SP_list[0][5] == '+':
+					ele_1 = SP_list[0]
+					ele_2 = SP_list[-1]
+				else:
+					ele_1 = [RLength-SP_list[-1][1], RLength-SP_list[-1][0]]+SP_list[-1][2:]
+					ele_2 = [RLength-SP_list[0][1],RLength-SP_list[0][0]]+SP_list[0][2:]
+				# print(ele_1)
+				# print(ele_2)
+				dis_ref = ele_2[2] - ele_1[3]
+				dis_read = ele_2[0] - ele_1[1]
+				if dis_ref < 100 and dis_read - dis_ref >= SV_size and dis_read - dis_ref <= 100000:
+					# print(min(ele_2[2], ele_1[3]), dis_read - dis_ref, read_name)
+					if ele_1[4] not in candidate['INS']:
+						candidate['INS'][ele_1[4]] = list()
+					candidate["INS"][ele_2[4]].append([min(ele_2[2], ele_1[3]), dis_read - dis_ref, read_name])
+
+				if dis_ref <= -SV_size:
+					if ele_2[4] not in candidate["DUP"]:
+						candidate["DUP"][ele_2[4]] = list()
+					candidate["DUP"][ele_2[4]].append([ele_2[2], ele_1[3], read_name])
 
 def acquire_clip_pos(deal_cigar):
 	seq = list(cigar.Cigar(deal_cigar).items())