def parse_fasta_and_write_new_file(results, contigs, output):
#pdb.set_trace()
for taxon, rows in results.iteritems():
outp = FastaWriter(os.path.join(output, "{}.fasta".format(taxon)))
inp = "{}.contigs.fasta".format(taxon.replace('_', '-'))
fasta_file = FastaReader(os.path.join(contigs, inp))
for fasta in fasta_file:
name = '_'.join(
fasta.identifier.lstrip('>').split('_')[:2]).lower()
if name in rows:
outp.write(fasta)
outp.close()
def main():
args = get_args()
pth = os.path.join(args.fastas, "*.fasta")
outf = FastaWriter(args.outfile)
conn = sqlite3.connect(args.db)
cur = conn.cursor()
counter = 0
for infile in glob.glob(pth):
sp = os.path.basename(infile).split('.')[0].replace('-', '_')
species = sp.replace('_', ' ').capitalize()
print "Working on {}".format(species)
partial = species.split(' ')[0].lower()[:3]
for read in FastaReader(infile):
# check for header match, if match get locus name for header
nn = read.identifier.split("_")[:2]
nn = "{}_{}".format(nn[0].strip('>').lower(), nn[1].lower())
query = "SELECT uce FROM match_map WHERE {0} = '{1}(+)' OR {0} = '{1}(-)'".format(
sp, nn)
cur.execute(query)
result = cur.fetchall()
#pdb.set_trace()
if result:
assert len(result) == 1, "More than 1 result"
#pdb.set_trace()
if args.fish:
uce = result[0][0].split('_')[0]
else:
uce = result[0][0]
read.identifier = """{3}{2} [organism={0}] [molecule=DNA] [moltype=genomic] [location=genomic] [note=ultra conserved element locus {1}] {0} ultra-conserved element locus {1}.""".format(
species, uce, partial, counter)
# write all to a common fasta
outf.write(read)
# if not match, pass
counter += 1
else:
pass
outf.close()
