- Timestamp:
- Dec 22, 2009, 8:12:09 AM (15 years ago)
- File:
-
- 1 edited
Legend:
- Unmodified
- Added
- Removed
-
liacs/dbdm/dbdm_4/parse-fasta.py
r55 r58 73 73 return(-1,'',head + tail) 74 74 75 # File names 76 file1 = "data/AE005174v2-1.fas" 77 file2 = "data/AE005174v2-2.fas" 75 78 76 79 # Get data 77 file1 = parse_file("data/AE005174v2-1.fas")78 file2 = parse_file("data/AE005174v2-2.fas")80 seq1 = parse_file(file1) 81 seq2 = parse_file(file1) 79 82 80 file1 = fasta.replace(file1)81 file2 = fasta.replace(file2)83 seq1 = fasta.replace(seq1) 84 seq2 = fasta.replace(seq2) 82 85 83 86 # Find overlap 84 (retval, common, result) = concat( file2,file1)87 (retval, common, result) = concat(seq2,seq1) 85 88 print retval, common 86 89 87 # Strictly speaking there is a gap of about 4 kbs (4000 bs) between file1 and 88 # file2, so lets' put that into the the statistics as well. Due to circular 90 print file1 91 stats(seq1) 92 print file1 93 stats(seq1) 94 95 # Strictly speaking there is a gap of about 4 kbs (4000 bs) between seq1 and 96 # seq2, so lets' put that into the the statistics as well. Due to circular 89 97 # nature, does not matter wether we add it in the beginning or in the end 98 print "Total (inc 4 kbs gap (n))" 90 99 result = result + "n" * 4000; 91 100 stats(result)
Note:
See TracChangeset
for help on using the changeset viewer.