def exonerate_pbsearch(fafile, esifile, espath, ep, threadnum, pblength, setpsize, dtm=10, minIdentity=75,
exonerateport=10005):
fa = loadfa.loadfa(fafile)
fa = fa.replace('n', '')
fa = fa.replace('N', '')
seqlen = len(fa)
windlen = 10000
sptime = int(seqlen/windlen)+1
pars = list()
#start exonerate server
# exonerate.start_exonerate_server(espath=espath, esifile=esifile, threadnum=24, exonerateport=exonerateport)
time.sleep(10)
for sc in range(sptime):
nowstart = sc * windlen
nowend = nowstart + windlen + 50
if nowend > seqlen:
nowend = seqlen-1
nowseq = fa[nowstart:nowend]
# print(nowseq)
par = dict()
par['sequence'] = nowseq
par['exonerateport'] = exonerateport
par['ep'] = ep
par['minIdentity'] = minIdentity
par['pblength'] = pblength
pars.append(par)
pool = Pool(threadnum)
res = pool.map(exonerate_getseqlen, pars)
probe = dict()
for i in res:
for j in i:
if j in probe:
probe[j] = +1
else:
probe[j] = 1
exonerate.stop_exonerate_server()
probelist = list()
print(probe)
for probeseq in probe:
if probe[probeseq] == 1:
probelist.append(probeseq)
return probelist
for nowseq in tmpseq:
if tmpseq[nowseq] == 1:
print('>',nowseq, sep='', file=tmpfaio)
print(nowseq, file=tmpfaio)
tmpfaio.close()
exonerate.exonerate_search_sequence2(ep=ep, sequencefile=tmpfa, exonerateport=exonerateport,
minIdentity=minIdentity)
return probelist
if __name__ == "__main__":
try:
prb = exonerate_pbsearch(fafile='/Users/Forrest/Documents/Project/Chorus/Test/PGSC_DM_v4.03_pseudomolecules_genes_exon_test.fa',
esifile='/Users/Forrest/Documents/Project/Chorus/Test/Gy14.sei',
espath='/opt/local/bin/exonerate-server', ep='/opt/local/bin/exonerate',
setpsize=5,
exonerateport=10005,
threadnum=2,
pblength = 45, dtm=10, minIdentity=75)
print(prb)
except KeyboardInterrupt:
exonerate.stop_exonerate_server()
sys.stderr.write("User interrupt\n")
sys.exit(0)