implemented barcode size selection

cribbslab · May 12, 2023 · 2438190 · 2438190
1 parent 52e4503
commit 2438190
Show file tree

Hide file tree

Showing 4 changed files with 13 additions and 6 deletions.
diff --git a/tallytrin/pipeline_10x.py b/tallytrin/pipeline_10x.py
@@ -177,7 +177,7 @@ def identify_bcumi(infile, outfile):
     PYTHON_ROOT = os.path.join(os.path.dirname(__file__), "python/")
 
     statement = '''python %(PYTHON_ROOT)s/10x_identify_barcode.py --outfile=%(outfile)s --infile=%(infile)s --whitelist=polyA_umi.dir/%(name)s.whitelist.txt
-                   --cmimode=%(cmimode)s'''
+                   --cmimode=%(cmimode)s --barcode=%(barcode)s'''
 
     P.run(statement, job_options='-t 24:00:00')
 
@@ -266,7 +266,7 @@ def mapping(infile, outfile):
 
     statement = '''minimap2  %(options)s %(cdna)s  %(infile)s > %(outfile)s 2> %(outfile)s.log'''
 
-    P.run(statement, job_options='-t 24:00:00')
+    P.run(statement, job_options='-t 24:00:00', job_threads=4)
 
 
 @transform(mapping,

diff --git a/tallytrin/pipeline_10x/pipeline.yml b/tallytrin/pipeline_10x/pipeline.yml
@@ -60,4 +60,8 @@ database:
 # this is so that the accuracy of the data can be measured: True 1, False 0
 cmi_mode: 1
 
-job_options: -t 48:00:00
+job_options: -t 48:00:00
+
+# Specify the barcode length you would like for 10x,
+# the default is 16 but for testing purposes this can be reduced
+barcode: 16
diff --git a/tallytrin/python/10x_identify_barcode.py b/tallytrin/python/10x_identify_barcode.py
@@ -1,3 +1,4 @@
+
 import sys
 import regex
 import cgatcore.iotools as iotools
@@ -29,6 +30,8 @@
                     help='name for output fastq files')
 parser.add_argument("--cmimode", default=None, type=str,
                     help='Run the script in cmi mode for accuracy evaluation')
+parser.add_argument("--barcode", default='16', type=int,
+                    help='Length of barcode required')
 
 args = parser.parse_args()
 
@@ -76,14 +79,14 @@
                 else:
                     umi = record.sequence[end_b:end_b+12]
                 if len(umi) == 15 and args.cmimode == '1':
-                    barcode = record.sequence[end_b+12:end_b+28]
+                    barcode = record.sequence[end_b+12:end_b+28][:int(args.barcode)]
                     barcodes.append(barcode)
                     seq_new = record.sequence[:begin_b]
                     quality_new = record.quality[:begin_b]
 
                     outfile.write("@%s\n%s\n+\n%s\n" % (record.name + "_" + barcode + "_" + umi, seq_new, quality_new))
                 else:
-                    barcode = record.sequence[end_b+12:end_b+28]
+                    barcode = record.sequence[end_b+12:end_b+28][:int(args.barcode)]
                     barcodes.append(barcode)
                     seq_new = record.sequence[:begin_b]
                     quality_new = record.quality[:begin_b]

diff --git a/tallytrin/python/correct_10xbarcode.py b/tallytrin/python/correct_10xbarcode.py
@@ -76,7 +76,7 @@ def closest_match(input_string, string_list):
 
         seq = record.sequence
         name = record.name.split("_")[0]
-        barcode = record.name.split("_")[1][2:12]
+        barcode = record.name.split("_")[1]
         umi = record.name.split("_")[2]
 
         if args.cmimode: