def collect_best_hits(filename):
d = {}
for n, record in enumerate(blastparser.parse_fp(open(filename))):
if n % 25000 == 0:
print >>sys.stderr, '...', filename, n
best_score = None
for hit in record.hits:
for match in hit.matches:
query = record.query_name
if query.startswith('gi'):
query = query.split('|', 2)[2]
subject = hit.subject_name
score = match.score
# only keep the best set of scores for any query
if best_score and best_score > score:
continue
best_score = score
x = d.get(query, [])
x.append((subject, score))
d[query] = x
if best_score and best_score != score:
break
return d
def collect_best_hits(filename, qfn=None):
d = {}
for n, record in enumerate(blastparser.parse_fp(open(filename))):
if n % 10000 == 0:
print '...', n
best_score = None
for hit in record.hits:
for match in hit.matches:
query = record.query_name
if qfn:
query = qfn(query)
subject = hit.subject_name
score = match.score
# only keep the best set of scores for any query
if best_score and best_score > score:
continue
best_score = score
x = d.get(query, [])
x.append((subject, score))
d[query] = x
if best_score and best_score != score:
break
return d
def read_blast(file_name):
hit_names=[]
for record in blastparser.parse_fp(open(file_name)):
for hit in record:
for match in hit.matches:
# output each match as a separate row
hit_names.append(record.query_name)
return hit_names
def read_blast(file_name):
hit_names={}
for record in blastparser.parse_fp(open(file_name)):
for hit in record:
for match in hit.matches:
# output each match as a separate row
# hit_names.append(hit.subject_name)
hit_names[record.query_name]=hit.subject_name
return hit_names
def collect_blast_hits_by_family(blastfile):
d = {}
for record in blastparser.parse_fp(open(blastfile)):
tr = record.query_name.split('.')[2]
assert tr.startswith('tr')
tr = int(tr[2:])
collect = []
for hit in record.hits:
for match in hit.matches:
if match.score >= BITSCORE_CUTOFF:
name = hit.subject_name.split('|')[1]
collect.append((name, match.score))
x = d.get(tr, [])
x.append(collect)
d[tr] = x
return d
def collect_blast_hits_by_family(blastfile):
d = {}
for record in blastparser.parse_fp(open(blastfile)):
tr = record.query_name.split('.')[2]
assert tr.startswith('tr')
tr = int(tr[2:])
collect = []
for hit in record.hits:
for match in hit.matches:
if match.score >= BITSCORE_CUTOFF:
name = hit.subject_name.split('|')[1]
collect.append((name, match.score))
x = d.get(tr, [])
x.append(collect)
d[tr] = x
return d
import screed
MIN_SCORE = 200
MIN_QUERY_LEN = int(sys.argv[3])
query_seqs = set([record.name for record in screed.open(sys.argv[4]) if len(record.sequence) >= MIN_QUERY_LEN])
covs = {}
for n, record in enumerate(screed.open(sys.argv[1])):
if n % 1000 == 0:
sys.stdout.write("+")
sys.stdout.flush()
covs[record.name] = [0] * len(record.sequence)
for n, record in enumerate(blastparser.parse_fp(open(sys.argv[2]))):
if n % 100 == 0:
sys.stdout.write(".")
sys.stdout.flush()
if record.query_name not in query_seqs:
continue
for hit in record.hits:
for match in hit.matches:
if match.score < MIN_SCORE:
continue
cov = covs.get(hit.subject_name)
if not cov:
continue
MIN_SCORE = 200
MIN_QUERY_LEN = int(sys.argv[3])
query_seqs = set([ record.name for record in screed.open(sys.argv[4]) \
if len(record.sequence) >= MIN_QUERY_LEN ])
covs = {}
for n, record in enumerate(screed.open(sys.argv[1])):
if n % 1000 == 0:
sys.stdout.write('+')
sys.stdout.flush()
covs[record.name] = [0] * len(record.sequence)
for n, record in enumerate(blastparser.parse_fp(open(sys.argv[2]))):
if n % 100 == 0:
sys.stdout.write('.')
sys.stdout.flush()
if record.query_name not in query_seqs:
continue
for hit in record.hits:
for match in hit.matches:
if match.score < MIN_SCORE:
continue
cov = covs.get(hit.subject_name)
if not cov:
continue
#! /usr/bin/env python
import sys
import csv
import blastparser
# get the filename as the first argument on the command line
filename = sys.argv[1]
# open it for reading
fp = open(filename)
# send output as comma-separated values to stdout
output = csv.writer(sys.stdout)
# parse BLAST records
for record in blastparser.parse_fp(fp):
for hit in record:
for match in hit.matches:
# output each match as a separate row
row = [record.query_name, hit.subject_name, match.score,
match.expect]
output.writerow(row)
ident = record.name
d[ident] = record.description
return d
# open the output file for reading
query_seqs = sys.argv[1]
against_seqs = sys.argv[2]
fp = open(sys.argv[3])
print >>sys.stderr, "reading query seq names from", query_seqs
query_db = load_names(query_seqs)
print >>sys.stderr, "reading against seq names from", against_seqs
against_db = load_names(against_seqs)
# send output as comma-separated values to stdout
output = csv.writer(sys.stdout)
# parse BLAST records
print >>sys.stderr, 'parsing BLAST output'
for record in blastparser.parse_fp(fp):
for hit in record:
for match in hit.matches:
query_descr = query_db.get(record.query_name, "")
against_descr = against_db.get(hit.subject_name, "")
# output each match as a separate row
row = [record.query_name, query_descr,
hit.subject_name, against_descr, match.score,
match.expect]
output.writerow(row)
#! /usr/bin/env python
import sys
sys.path.insert(0, '/u/t/dev/blastkit/lib')
import blastparser
import screed
from pygr.sequence import Sequence
seqsfile = sys.argv[1]
genome_name = seqsfile[:-3]
seqdb = screed.ScreedDB(seqsfile)
blastfile = 'large.x.' + genome_name
for record in blastparser.parse_fp(open(blastfile)):
tagname = record.query_name ##
for hit in record.hits:
for match in hit:
seq = Sequence(seqdb[hit.subject_name].sequence, tagname)
start, end = match.subject_start, match.subject_end
subseq = seq[start:end]
print '>%s.%s\n%s' % (genome_name, tagname, subseq)
break
break
#! /usr/bin/env python
import sys
import blastparser
import screed
MIN_SCORE = 200
covs = {}
for record in screed.open(sys.argv[1]):
covs[record.name] = [0] * len(record.sequence)
#print 'before'
for record in blastparser.parse_fp(open(sys.argv[2])):
#print record
#print 'test'
#sys.stdout.write('.')
#sys.stdout.flush()
#print record
for hit in record.hits:
#print hit
for match in hit.matches:
cov = covs[hit.subject_name]
start = min(match.subject_start, match.subject_end) - 1
end = max(match.subject_start, match.subject_end)
for i in range(start, end):
cov[i] = 1
print ''
import screed
import blastparser
import sys
infile = sys.argv[1]
blast_file = sys.argv[0]
lengths={}
for n, record in enumerate(screed.open(infile)):
lengths[record['name']]=len(record['sequence'])
idn, cnt, num = 0,0,0
for hits in blastparser.parse_fp(blast_file):
hit_len=0
hit_idn=0
for match in hits[0].matches:
hit_len=len(match.query_sequence)+hit_len
hit_idn=match.identity+hit_idn
hit_idn=hit_idn/float(hits[0])
covered = hit_len/float(lengths[hits.query_name])
if covered >= 1:
idn = hit_idn + idn
else:
idn = hit_idn*covered +idn
#! /usr/bin/env python
import sys
import blastparser
import screed
MIN_SCORE=200
covs = {}
for record in screed.open(sys.argv[1]):
covs[record.name] = [0] * len(record.sequence)
#print 'before'
for record in blastparser.parse_fp(open(sys.argv[2])):
#print record
#print 'test'
#sys.stdout.write('.')
#sys.stdout.flush()
#print record
for hit in record.hits:
#print hit
for match in hit.matches:
cov = covs[hit.subject_name]
start = min(match.subject_start, match.subject_end) - 1
end = max(match.subject_start, match.subject_end)
for i in range(start, end):
cov[i] = 1
fa_1 = sys.argv[1]
fa_2 = sys.argv[2]
blast_1x2 = sys.argv[3]
blast_2x1 = sys.argv[4]
seqs1 = {}
seqs2 = {}
for record in screed.open(fa_1):
seqs1[record.name] = record.sequence
for record in screed.open(fa_2):
seqs2[record.name] = record.sequence
for b in blastparser.parse_fp(open(blast_1x2)):
assert b.query_name in seqs1
del seqs1[b.query_name]
for b in blastparser.parse_fp(open(blast_2x1)):
assert b.query_name in seqs2
del seqs2[b.query_name]
print len(seqs1), 'missing 1'
print len(seqs2), 'missing 2'
fp = open(blast_1x2 + '.missing', 'w')
for name in seqs1:
fp.write(">%s\n%s\n" % (name, seqs1[name]))
fp.close()