added --split-intervals option, fixes #189

CGATOxford · Jun 9, 2015 · baf0a13 · baf0a13
1 parent d1662a5
commit baf0a13
Show file tree

Hide file tree

Showing 3 changed files with 24 additions and 2 deletions.
diff --git a/scripts/bam_vs_bed.py b/scripts/bam_vs_bed.py
@@ -44,7 +44,7 @@
 
 Type::
 
-   cgat bam_vs_bed BAM BED [OPTIONS] 
+   cgat bam_vs_bed BAM BED [OPTIONS]
    cgat bam_vs_bed --bam-file=BAM --bed-file=BED [OPTIONS]
 
 where BAM is either a bam or bed file and BED is a bed file.
@@ -102,11 +102,20 @@ def main(argv=None):
         "processing. By default the file is assumed to be sorted "
         "[%default]")
 
+    parser.add_option(
+        "--split-intervals", dest="split_intervals",
+        action="store_true",
+        help="treat split BAM intervals, for example spliced intervals, "
+        "as separate intervals. Note that a single alignment might be "
+        "counted several times as a result. "
+        "[%default]")
+
     parser.set_defaults(
         min_overlap=0.5,
         filename_bam=None,
         filename_bed=None,
         sort_bed=False,
+        split_intervals=False,
     )
 
     # add common options (-h/--help, ...) and parse command line
@@ -199,10 +208,16 @@ def main(argv=None):
     else:
         bedcmd = filename_bed
 
+    if options.split_intervals:
+        split = "-split"
+    else:
+        split = ""
+
     # IMS: newer versions of intersectBed have a very high memory
     #      requirement unless passed sorted bed files.
-    statement = """intersectBed %(format)s %(filename_bam)s
+    statement = """bedtools intersect %(format)s %(filename_bam)s
     -b %(bedcmd)s
+    %(split)s
     -sorted -bed -wo -f %(min_overlap)f""" % locals()
 
     E.info("starting counting process: %s" % statement)

diff --git a/tests/bam_vs_bed.py/tests.yaml b/tests/bam_vs_bed.py/tests.yaml
@@ -17,5 +17,11 @@ paired_unsorted:
         references: [paired.tsv]
         options: --sort-bed <DIR>/paired.bam <DIR>/context.paired_unsorted.bed.gz
 
+paired_split:
+        stdin: null
+        outputs: [stdout]
+        references: [paired.tsv]
+        options: --split-intervals <DIR>/paired.bam <DIR>/context.paired.bed.gz
+
 
 
diff --git a/tests/option_list.tsv b/tests/option_list.tsv
@@ -880,6 +880,7 @@ split-at-tag	ok			scripts/farm.py,scripts/nofarm.py
 split-exons	ok			scripts/align_transcripts.py
 split-fasta-identifier	ok			scripts/fasta2table.py
 split-fields	ok			scripts/table2table.py
+split-intervals	ok			scripts/bam_vs_bed.py
 spp-srange-max	ok			scripts/runSPP.py
 spp-srange-min	ok			scripts/runSPP.py
 spp-z-threshold	ok			scripts/runSPP.py