Changeset 55 for liacs/dbdm/dbdm_4/parse-fasta.py

Timestamp:

Dec 22, 2009, 7:52:22 AM (15 years ago)

Author:

Rick van der Zwet

Message:

Make it 'production' ready

File:

• liacs/dbdm/dbdm_4/parse-fasta.py (modified) (2 diffs)

liacs/dbdm/dbdm_4/parse-fasta.py

@@ -1 +1 @@
 #!/usr/bin/env python
+#
+# Parse 2 FASTA files and print statistics
+# BSDLicence
+# Rick van der Zwet - 0433373 - <info@rickvaderzwet.nl>
 from Bio import SeqIO,Seq
 from Bio import Alphabet
@@ -70 +74 @@
 
 
+# Get data
 file1 = parse_file("data/AE005174v2-1.fas")
 file2 = parse_file("data/AE005174v2-2.fas")
+
 file1 = fasta.replace(file1)
 file2 = fasta.replace(file2)
+
+# Find overlap
 (retval, common, result) = concat(file2,file1)
 print retval, common
+
+# Strictly speaking there is a gap of about 4 kbs (4000 bs) between file1 and
+# file2, so lets' put that into the the statistics as well. Due to circular
+# nature, does not matter wether we add it in the beginning or in the end
+result = result + "n" * 4000;
 stats(result)
 
+# Write to file for later further processing
 out = open("full_contig.raw","w")
 out.write(result)