def main(argv=None):
if argv is None:
argv = sys.argv
ids = None
intIds = None
blastfileGlob = None
fastafileGlob = None
outputfile = None
try:
optlist, args = getopt(argv[1:], "hb:f:o:i:")
except:
print "Illegal option"
print ""
print HELP_STRING
sys.exit(1)
for (opt, opt_arg) in optlist:
if opt == "-h":
print ""
print HELP_STRING
sys.exit(1)
elif opt == "-i":
ids = opt_arg.split(",")
intIds = [int(i) for i in ids]
elif opt == "-b":
blastfileGlob = glob(opt_arg)
elif opt == "-f":
fastafileGlob = glob(opt_arg)
elif opt == "-o":
outputfile = opt_arg
if not blastfileGlob or not fastafileGlob or not outputfile or not intIds:
print "You must include the ids to retrieve (-i), blast file (-b), input fasta file (-f) and the output filename (-o)"
print blastfileGlob, fastafileGlob, outputfile, intIds
sys.exit(1)
#print intIds
pipelineUtils.getFastaForTaxonomy(intIds, blastfileGlob, fastafileGlob, outputfile)
def doPostProcess(blastResults, fastaInput):
rootName, extension = os.path.splitext(blastResults)
bestHitsName = rootName + ".bestHits.bln"
bestHitsName2 = rootName + ".besthits.bln"
dp = rootName + ".bubble_"
reportName = rootName + ".tax_"
# do best hits
print >> sys.stderr, 'STEP 5: calculate besthits file'
print >> sys.stderr, ' : besthits filename: >', bestHitsName, "<"
print >> sys.stderr, ' : besthits filename: >', bestHitsName2, "<"
#updated 2015-08-04 so to stop stupid bug from not seeing ".besthits.bln" files, like when copied from DOS
if os.path.exists(bestHitsName2):
bestHitsName = bestHitsName2
print >> sys.stderr, ' : found besthits filename: >', bestHitsName2, "<"
if not os.path.exists(bestHitsName):
blastUtils.keepAllBestHits(blastResults, bestHitsName)
else:
print >> sys.stderr, '<skip>: besthits file already exists'
# report taxonomy
reportTaxonomyJWC(bestHitsName,
printTargets=True,
outputPrefix=reportName,
dotPrefix=dp)
if fastaInput:
pipelineUtils.getFastaForTaxonomy([10239], bestHitsName, fastaInput,
rootName + "_viral.fa")
pipelineUtils.getFastaForTaxonomy([10292], bestHitsName, fastaInput,
rootName + "_herp.fa")
pipelineUtils.getFastaForTaxonomy([11632], bestHitsName, fastaInput,
rootName + "_retro.fa")
def doPostProcess(blastResults, fastaInput):
rootName, extension = os.path.splitext(blastResults)
bestHitsName = rootName + ".bestHits.bln"
bestHitsName2 = rootName + ".besthits.bln"
dp = rootName + ".bubble_"
reportName = rootName + ".tax_"
# do best hits
print >> sys.stderr, 'STEP 5: calculate besthits file'
print >> sys.stderr, ' : besthits filename: >', bestHitsName, "<"
print >> sys.stderr, ' : besthits filename: >', bestHitsName2, "<"
#updated 2015-08-04 so to stop stupid bug from not seeing ".besthits.bln" files, like when copied from DOS
if os.path.exists(bestHitsName2):
bestHitsName = bestHitsName2
print >> sys.stderr, ' : found besthits filename: >', bestHitsName2, "<"
if not os.path.exists(bestHitsName):
blastUtils.keepAllBestHits(blastResults, bestHitsName)
else:
print >> sys.stderr, '<skip>: besthits file already exists'
# report taxonomy
reportTaxonomyJWC(bestHitsName, printTargets=True, outputPrefix=reportName, dotPrefix=dp)
if fastaInput:
pipelineUtils.getFastaForTaxonomy([10239], bestHitsName, fastaInput, rootName + "_viral.fa")
pipelineUtils.getFastaForTaxonomy([10292], bestHitsName, fastaInput, rootName + "_herp.fa")
pipelineUtils.getFastaForTaxonomy([11632], bestHitsName, fastaInput, rootName + "_retro.fa")
def doPostProcess(blastResults, fastaInput):
rootName, extension = os.path.splitext(blastResults)
bestHitsName = rootName + ".bestHits.bln"
dp = rootName + ".bubble_"
reportName = rootName + ".tax_"
# do best hits
blastUtils.keepAllBestHits(blastResults, bestHitsName)
# report taxonomy
pipelineUtils.reportTaxonomy(bestHitsName,
printTargets=True,
outputPrefix=reportName,
dotPrefix=dp)
if fastaInput:
pipelineUtils.getFastaForTaxonomy([10239], bestHitsName, fastaInput,
rootName + "_viral.fa")
pipelineUtils.getFastaForTaxonomy([10292], bestHitsName, fastaInput,
rootName + "_herp.fa")
pipelineUtils.getFastaForTaxonomy([11632], bestHitsName, fastaInput,
rootName + "_retro.fa")