def parseHmmerResults(self, fileName, resultsManager, bSkipAdjCorrection):
"""Parse HMMER results."""
try:
with open(fileName, 'r') as hmmerHandle:
try:
HP = HMMERParser(hmmerHandle)
except:
print("Error opening HMM file: ", fileName)
raise
while True:
hit = HP.next()
if hit is None:
break
resultsManager.addHit(hit)
# retain only best hit to PFAM clan
pfam = PFAM(DefaultValues.PFAM_CLAN_FILE)
resultsManager.markerHits = pfam.filterHitsFromSameClan(resultsManager.markerHits)
# correct for errors in ORF calling
if not bSkipAdjCorrection:
resultsManager.identifyAdjacentMarkerGenes()
except IOError as detail:
sys.stderr.write(str(detail) + "\n")
def __createMarkerHMMs(self, binMarkerSet, outputFile, bReportProgress=True):
"""Create HMM file for markers."""
# get list of marker genes
markerGenes = binMarkerSet.getMarkerGenes()
# get all genes from the same clan as any marker gene
pfam = PFAM(DefaultValues.PFAM_CLAN_FILE)
genesInSameClan = pfam.genesInSameClan(markerGenes)
# extract marker genes along with all genes from the same clan
allMarkers = markerGenes | genesInSameClan
if bReportProgress:
self.logger.info(" There are %d genes in the marker set and %d genes from the same PFAM clan." % (len(markerGenes), len(genesInSameClan)))
# create file with all model accession numbers
keyFile = os.path.join(tempfile.gettempdir(), str(uuid.uuid4()))
fout = open(keyFile, 'w')
for modelAcc in allMarkers:
fout.write(modelAcc + '\n')
fout.close()
# fetch specified models
HF = HMMERRunner(mode='fetch')
HF.fetch(DefaultValues.HMM_MODELS, keyFile, outputFile, bKeyFile=True)
# index the HMM file
if os.path.exists(outputFile + '.ssi'):
os.remove(outputFile + '.ssi')
HF.index(outputFile)
# remove key file
os.remove(keyFile)
def parseHmmerResults(self, fileName, resultsManager, bSkipAdjCorrection):
"""Parse HMMER results."""
try:
with open(fileName, 'r') as hmmerHandle:
try:
HP = HMMERParser(hmmerHandle)
except:
print("Error opening HMM file: ", fileName)
raise
while True:
hit = HP.next()
if hit is None:
break
resultsManager.addHit(hit)
# retain only best hit to PFAM clan
pfam = PFAM(DefaultValues.PFAM_CLAN_FILE)
resultsManager.markerHits = pfam.filterHitsFromSameClan(
resultsManager.markerHits)
# correct for errors in ORF calling
if not bSkipAdjCorrection:
resultsManager.identifyAdjacentMarkerGenes()
except IOError as detail:
sys.stderr.write(str(detail) + "\n")
def __createMarkerHMMs(self, binMarkerSet, outputFile, bReportProgress=True):
"""Create HMM file for markers."""
# get list of marker genes
markerGenes = binMarkerSet.getMarkerGenes()
# get all genes from the same clan as any marker gene
pfam = PFAM(DefaultValues.PFAM_CLAN_FILE)
genesInSameClan = pfam.genesInSameClan(markerGenes)
# extract marker genes along with all genes from the same clan
allMarkers = markerGenes | genesInSameClan
if bReportProgress:
self.logger.info(" There are %d genes in the marker set and %d genes from the same PFAM clan." % (len(markerGenes), len(genesInSameClan)))
# create file with all model accession numbers
keyFile = os.path.join(tempfile.gettempdir(), str(uuid.uuid4()))
fout = open(keyFile, 'w')
for modelAcc in allMarkers:
fout.write(modelAcc + '\n')
fout.close()
# fetch specified models
HF = HMMERRunner(mode='fetch')
HF.fetch(DefaultValues.HMM_MODELS, keyFile, outputFile, bKeyFile=True)
# index the HMM file
if os.path.exists(outputFile + '.ssi'):
os.remove(outputFile + '.ssi')
HF.index(outputFile)
# remove key file
os.remove(keyFile)