Python for a BLASTed file

Question

[Kevin_Smith]

I have a BLASTED file as tabular with comment lines. I have to find the total hits. I will appreciate any help for a python script.

[piet]

Please make yourself familiar with the EXACT meaning of the columns in your blast output file. The main problem with blast output is to decide if a hit (a row in your table) is relevant for your project. For example, the second row in your table indicates a short sequence repeat within 'sequence_A'. Is this relevant or not?

Answer 1

[Andrzej Zielezinski]

Here's your Python script:

from itertools import groupby

fh = open('BLASTED.txt')
oh = open('BLASTED.txt.out', 'w')
queries_no = 0
hits_no = 0
for qid, hsps in groupby(fh, lambda l: l.split()[0]):
    if qid.startswith('#'): continue
    hits = len(set([l.split()[1] for l in hsps]))
    hits_no += hits
    queries_no += 1
    oh.write('{0}\t{1}\n'.format(qid, hits))
oh.close()
fh.close()

print 'Total queries  :', queries_no
print 'Total hits     :', hits_no
print 'Averaged hits  :', float(hits_no)/queries_no

Answer 2

[Michael]

The average number of hits is of very limited value, especially if the number of hits to display was restricted in the blast search.

To do this in practice on a tabular output is very simple: each line represents a hit, therefore you can count the occurrences of each unique query id (sequence_A := 5) and divide by the total number of query ids (1).

Answer 3

[Kevin_Smith]

Thank you very much a.zielezinski. The script works perfect !!

Answer 4

[Kevin_Smith]

I have to consider one hit per subject sequence. All of this using the BLASTED txt file.