def _psipred(self):
"""
Run psipred on the sequence.
"""
debug("Running psipred prediction")
from hpf.seq import TemporaryRecordFile
from hpf.pdb.psipred import Psipred, PsipredOptions
from Bio import SeqIO
fasta = os.path.join(self.scratch,self.prediction_code,self.prediction_code+".fasta")
with open(fasta,"w") as handle:
SeqIO.write([self.sequence.record], handle, "fasta")
psipred = self.psipred
from Bio.Blast.NCBIStandalone import blastpgp
chk = fasta+".chk"
import subprocess
cmd = subprocess.Popen(["which", "blastpgp"], stdout=subprocess.PIPE).communicate()[0].strip()
debug("Using",cmd)
result,error = blastpgp(cmd,
"nr",
fasta,
npasses=3,
checkpoint_outfile=chk,
expectation=1e-4,
model_threshold=1e-4,
align_outfile="/dev/null")
debug(result.readlines())
debug(error.readlines())
options = PsipredOptions(fasta,
profile=chk,
output=psipred+".1",
output2=psipred+".2",
horiz=psipred,
cwd = os.path.join(self.scratch,self.prediction_code))
prediction = Psipred(options).run()
db_pred = PsipredFactory().create(prediction,sequence_key=self.sequence.id)
self.session.add(db_pred)
self.session.commit()
#assert os.path.exists(psipred)
return psipred
def run(self, ):
"""
Cobbles together elements for running Psipred (writes fasta, runs blast to get checkpoint,
runs psipred and psipass2, and then parses results and uploads to DB if set to).
IF dbstore is true, adds Psipred ORM object to hpf database and sets self.dbo.
RETURNS hpf.pdb.psipred.PsipredPrediction object
"""
import subprocess
from Bio import SeqIO
from Bio.Blast.NCBIStandalone import blastpgp
from hpf.hddb.db import Psipred as PsipredORM, PsipredFactory
from sqlalchemy.exc import IntegrityError
# Write fasta file
if self.debug: print "Psipred: writing fasta file"
with open(self.fasta_file, 'w') as handle:
SeqIO.write([self.sequence.record], handle, "fasta")
# Get exe path and run blast
if self.debug: print "Psipred: running blastpgp against '{0}' DB to create checkpoint file".format(self.nr_db)
blast_cmd = subprocess.Popen(["which", "blastpgp"], stdout=subprocess.PIPE).communicate()[0].strip()
result,error = blastpgp(blastcmd=blast_cmd,
program='blastpgp',
database=self.nr_db,
infile=self.fasta_file,
npasses=3,
checkpoint_outfile=self.chkpt_file,
expectation=1e-4,
model_threshold=1e-4,
align_outfile="/dev/null")
# Note: must call something that blocks on blastpgp results (need to wait for cmd to finish)
res = result.read()
err = error.read()
if self.debug:
print "Result: ", res
print "Error/Warning: ", err
# Create Psipred options object and run Psipred 3.2 on them
if self.debug: print "Psipred: running Psipred 3.2"
options = PsipredOptions(self.fasta_file,
profile=self.chkpt_file,
output=self.psipred_file+".1",
output2=self.psipred_file+".2",
horiz=self.psipred_file,
cwd=self.dir)
self.prediction = Psipred32(options).run()
# Add to database (optional) and return
if self.dbstore:
if self.debug: print "Psipred: adding psipred DBO to hpf database"
psipred_dbo = PsipredFactory().create(self.prediction, sequence_key=self.sequence.id, ginzu_version=self.ginzu_version)
self.session.add(psipred_dbo)
try:
self.session.commit()
except IntegrityError:
print "Psipred entry for <seq {0}, ginzu_version {1}> already exists in DB. Returning existing object".format(self.sequence_key, self.ginzu_version)
self.session.rollback()
psipred_dbo = self.session.query(PsipredORM).filter_by(sequence_key=self.sequence.id, ginzu_version=self.ginzu_version).first()
self.session.refresh(psipred_dbo)
self.dbo = psipred_dbo
return self.prediction