add improper_check_maximum_unique_pairs option

genomon_sv/arg_parser.py

@@ -9,7 +9,7 @@ def create_parser():
     # top level parser
     parser = argparse.ArgumentParser(prog = "GenomonSV", formatter_class=argparse.ArgumentDefaultsHelpFormatter)
 
-    parser.add_argument("--version", action = "version", version = "GenomonSV-0.6.0")
+    parser.add_argument("--version", action = "version", version = "GenomonSV-0.6.1b1")
 
     subparsers = parser.add_subparsers()
 
@@ -82,6 +82,9 @@ def create_parser():
     cluster_improper_group.add_argument("--improper_check_margin_size", type = int, default = 1500,
                                         help = "This should be sufficiently greater than insert size, but the computational time will increase when too large (default: %(default)s)")
 
+    cluster_improper_group.add_argument("--improper_check_maximum_unique_pairs", type = int, default = 10000,
+                                        help = "The number of unique improper pairs at each local region. \
+                                                Set mainly for avoiding local regions with extremely high depths and huge amount of junction reads (default: %(default)s)")
 
     parse_parser.set_defaults(func = genomonSV_parse)
     ####################   

genomon_sv/parseFunction.py

@@ -570,7 +570,7 @@ def clusterJunction(inputFilePath, outputFilePath, check_margin_size, maximum_un
 
         if len(mergedJunction) > maximum_unique_pairs:
             print >> sys.stderr, "Exceeded maximum number of unique junction pairs at %s:%s-%s" % (F[0], F[1], F[2])
-            print >> sys.stderr, "Skipp %s:%s-%s" % (F[0], F[1], str(int(F[2]) + check_margin_size))
+            print >> sys.stderr, "Skip %s:%s-%s" % (F[0], F[1], str(int(F[2]) + check_margin_size))
             mergedJunction = {}
             mergedBedpeInfo = {}
             skip_pos = int(F[1]) + check_margin_size
@@ -737,17 +737,21 @@ def makeImproperBedpe(inputFilePath, outputFilePath, junction_dist_margin, clipp
 
 
 
-def clusterImproperBedpe(inputFilePath, outputFilePath, check_margin_size):
+def clusterImproperBedpe(inputFilePath, outputFilePath, check_margin_size, maximum_unique_pairs):
 
     ####################
     # cluster and summarize improper read pair bed file
     hIN = open(inputFilePath, "r")
     hOUT = open(outputFilePath, "w")
 
     mergedBedpe = {}
+    temp_chr = None
+    skip_pos = 0
     for line in hIN:
 
         F = line.rstrip('\n').split('\t')
+        if F[0] != temp_chr: temp_chr, skip_pos = F[0], 0
+        if int(F[1]) < skip_pos: continue
 
         match = 0
         delList = []
@@ -796,6 +800,11 @@ def clusterImproperBedpe(inputFilePath, outputFilePath, check_margin_size):
             newKey = '\t'.join([F[0], F[1], F[2], F[3], F[4], F[5], F[8], F[9]])
             mergedBedpe[newKey] = F[6] + '\t' + F[7] + '\t' + F[10]
 
+        if len(mergedBedpe) > maximum_unique_pairs:
+            print >> sys.stderr, "Exceeded maximum number of unique improper pairs at %s:%s-%s" % (F[0], F[1], F[2])
+            print >> sys.stderr, "Skip %s:%s-%s" % (F[0], F[1], str(int(F[2]) + check_margin_size))
+            mergedBedpe
+            skip_pos = int(F[1]) + check_margin_size
 
     for key in sorted(mergedBedpe):
 

genomon_sv/run.py

@@ -84,7 +84,8 @@ def genomonSV_parse(args):
 
     # cluster read pairs possibly representing the same junction
     utils.processingMessage("clustering improperly aligned read pairs")
-    parseFunction.clusterImproperBedpe(args.output_prefix + ".improper.bedpe", args.output_prefix + ".improper.clustered.unsort.bedpe", args.improper_check_margin_size)
+    parseFunction.clusterImproperBedpe(args.output_prefix + ".improper.bedpe", args.output_prefix + ".improper.clustered.unsort.bedpe", 
+                                       args.improper_check_margin_size, args.improper_check_maximum_unique_pairs)
 
     utils.processingMessage("sorting clustered improperly aligned read pairs")
     utils.sortBedpe(args.output_prefix + ".improper.clustered.unsort.bedpe", args.output_prefix + ".improper.clustered.bedpe")

setup.py

@@ -11,7 +11,7 @@
 
 setup(
     name = 'genomon_sv',
-    version = '0.6.0',
+    version = '0.6.1b1',
     description='Python tools for detecting somatic structural variation from cancer genome sequencing data.',
     url = 'https://github.com/friend1ws/GenomonSV',
     author = 'Yuichi Shiraishi',