Add scripts for pre-processing data

utils/preprocessBAMstLFR.py

@@ -0,0 +1,16 @@
+import pysam
+import sys
+import os
+
+bamFile = pysam.AlignmentFile(sys.argv[1], "rb");
+outFile = pysam.AlignmentFile(os.path.splitext(sys.argv[1])[0] + "_barcodes_extracted" + os.path.splitext(sys.argv[1])[1], "wb", template=bamFile)
+
+iter = bamFile.fetch()
+for al in iter:
+        t = al.query_name.split("#")
+        al.query_name = t[0]
+        al.set_tag("BX", t[1])
+        outFile.write(al)
+
+bamFile.close()
+outFile.close()

utils/preprocessFASTQstLFR.py

@@ -0,0 +1,27 @@
+import sys
+import re
+import csv
+import os
+
+r = open(sys.argv[1]);
+o = open(os.path.splitext(sys.argv[1])[0] + "_barcodes_in_headers" + os.path.splitext(sys.argv[1])[1], "w");
+
+header = r.readline()[:-1];
+while header != '':
+	line = r.readline();
+	data = line;
+	line = r.readline();
+	data += line;
+	line = r.readline();
+	data += line;
+
+	t = header.split("#");
+	tt = t[1].split("/");
+	barcode = tt[0]
+	o.write(t[0] + "/" + tt[1].split("\t")[0].split(" ")[0] + "\tBX:Z:" + barcode + "\n");
+	o.write(data);
+
+	header = r.readline()[:-1]
+
+r.close();
+o.close();

utils/preprocessTELL-Seq.py

@@ -0,0 +1,48 @@
+import sys
+import re
+import csv
+import os
+
+r1 = open(sys.argv[1]);
+r2 = open(sys.argv[2]);
+i1 = open(sys.argv[3]);
+
+o1 = open(os.path.splitext(sys.argv[1])[0] + "_barcodes_in_headers" + os.path.splitext(sys.argv[1])[1], "w");
+o2 = open(os.path.splitext(sys.argv[2])[0] + "_barcodes_in_headers" + os.path.splitext(sys.argv[2])[1], "w");
+
+header1 = r1.readline()[:-1];
+while header1 != '':
+	line = r1.readline();
+	data1 = line;
+	line = r1.readline();
+	data1 += line;
+	line = r1.readline();
+	data1 += line;
+
+	header2 = r2.readline()[:-1];
+	line = r2.readline();
+	data2 = line;
+	line = r2.readline();
+	data2 += line;
+	line = r2.readline();
+	data2 += line;
+
+	header3 = i1.readline();
+	barcode = i1.readline();
+	i1.readline();
+	i1.readline();
+
+	h = header1.split(" ")[0].split("\t")[0];
+	o1.write(h + "\tBX:Z:" + barcode);
+	o1.write(data1);
+	h = header2.split(" ")[0].split("\t")[0];
+	o2.write(h + "\tBX:Z:" + barcode);
+	o2.write(data2);
+
+	header1 = r1.readline()[:-1]
+
+r1.close();
+r2.close();
+i1.close();
+o1.close();
+o2.close();