def parseArgs():
parser = argparse.ArgumentParser(description=USAGE, \
formatter_class=argparse.RawDescriptionHelpFormatter)
parser.add_argument("reads", metavar="reads", type=str, \
help="Input reads .fasta or .fastq")
parser.add_argument("-t", "--target", type=str, \
help="Target sequence name")
parser.add_argument("-T", "--Target", type=str, \
help="Fasta file containing target sequence")
parser.add_argument("-s", "--super", dest="super", action="store_true",\
help="Treat each read as the target once")
parser.add_argument("-m", "--maxtail", type = int, default=sys.maxint, \
help="Max number of bases allowed to be in tail (inf)")
parser.add_argument("-n", "--nproc", dest="nproc", default=1, type=int,\
help="Number of processors to use with blasr (1)")
parser.add_argument("-o", "--outname", dest="outname", default="polish.out", \
type=str, \
help="Base name for output files (polish.out)")
parser.add_argument("--debug", action="store_true")
args = parser.parse_args()
setupLogging(args.debug)
#I don't think this is exhaustive
if (args.target is not None and args.Target is not None) \
or (args.super and (args.target is not None or args.Target is not None)):
print "Error! only specify one of --super or --target or --Target"
exit(1)
return args
def parseArgs():
"""
input dir
predicted gapsize
if argument says that we need to extract the seeds we will have a single paramters
extractFlanks
"""
parser = argparse.ArgumentParser(description=USAGE, \
formatter_class=argparse.RawDescriptionHelpFormatter)
parser.add_argument("asmdir", metavar="DIR", type=str, \
help="Local assembly directory for a gap")
parser.add_argument("-t", "--maxTrim", type=int, default=100, \
help="Maxmum trim allowed (100)")
parser.add_argument("-w", "--maxWiggle", type=int, default=400, \
help="Maxmum wiggle for gap spanning allowed (400)")
parser.add_argument("-p", "--predictedGapSize", type=int, default=None)
parser.add_argument("-n", "--nproc", type=int, default=1)
parser.add_argument("-k", "--keepTemp", action="store_true",\
help="Keep temporary files")
parser.add_argument("--tempDir", type=str, default=None,
help="Where to write temporary files (DIR)")
parser.add_argument("--debug", action="store_true")
args = parser.parse_args()
if args.asmdir.endswith("/"):
args.asmdir = args.asmdir[:-1]
if args.tempDir is None:
args.tempDir = args.asmdir
setupLogging(args.debug)
return args
def parseArgs(argv):
parser = argparse.ArgumentParser(description=USAGE, \
formatter_class=argparse.RawDescriptionHelpFormatter)
parser.add_argument("m4", metavar="M4", type=str, \
help="M4 containing mapped reads' alignments")
parser.add_argument("reads", metavar="READS", type=str,\
help="Fasta/Fastq containing reads' sequence")
parser.add_argument("ref", metavar="REFERENCE", type=str,\
help="REFERENCE to map tails to")
parser.add_argument("-t", "--minTail", type=int, default=100,\
help="Minimum tail length to attempt remapping (100)")
parser.add_argument("-n", "--nproc", type=int, default=1,\
help="Number of processors to use (1)")
parser.add_argument("-o", "--output", type=str, default=None, \
help="Output Name (M4.tails.m4)")
parser.add_argument("-i", "--inplace", action="store_true", \
help="Append the results to the input m4 file. Overrules --output")
parser.add_argument("--noSa", action="store_true", \
help="Don't use reference's sa")
parser.add_argument("--temp",
type=str,
default=tempfile.gettempdir(),
help="Where to save temporary files")
parser.add_argument("--debug", action="store_true")
args = parser.parse_args(argv)
if args.inplace:
args.output = args.m4
elif args.output is None:
args.output = args.m4[:-3] + ".tails.m4"
setupLogging(args.debug)
return args
def parseArgs(argv):
parser = argparse.ArgumentParser(description=USAGE, \
formatter_class=argparse.RawDescriptionHelpFormatter)
parser.add_argument("m4", metavar="M4", type=str, \
help="M4 containing mapped reads' alignments")
parser.add_argument("reads", metavar="READS", type=str,\
help="Fasta/Fastq containing reads' sequence")
parser.add_argument("ref", metavar="REFERENCE", type=str,\
help="REFERENCE to map tails to")
parser.add_argument("-t", "--minTail", type=int, default=100,\
help="Minimum tail length to attempt remapping (100)")
parser.add_argument("-n", "--nproc", type=int, default=1,\
help="Number of processors to use (1)")
parser.add_argument("-o", "--output", type=str, default=None, \
help="Output Name (M4.tails.m4)")
parser.add_argument("-i", "--inplace", action="store_true", \
help="Append the results to the input m4 file. Overrules --output")
parser.add_argument("--noSa", action="store_true", \
help="Don't use reference's sa")
parser.add_argument("--temp", type=str, default=tempfile.gettempdir(),
help="Where to save temporary files")
parser.add_argument("--debug", action="store_true")
args = parser.parse_args(argv)
if args.inplace:
args.output = args.m4
elif args.output is None:
args.output = args.m4[:-3] + ".tails.m4"
setupLogging(args.debug)
return args
def parseArgs():
"""
input dir
predicted gapsize
if argument says that we need to extract the seeds we will have a single paramters
extractFlanks
"""
parser = argparse.ArgumentParser(description=USAGE, \
formatter_class=argparse.RawDescriptionHelpFormatter)
parser.add_argument("asmdir", metavar="DIR", type=str, \
help="Local assembly directory for a gap")
parser.add_argument("-t", "--maxTrim", type=int, default=100, \
help="Maxmum trim allowed (100)")
parser.add_argument("-w", "--maxWiggle", type=int, default=400, \
help="Maxmum wiggle for gap spanning allowed (400)")
parser.add_argument("-p", "--predictedGapSize", type=int, default=None)
parser.add_argument("-n", "--nproc", type=int, default=1)
parser.add_argument("-k", "--keepTemp", action="store_true",\
help="Keep temporary files")
parser.add_argument("--tempDir", type=str, default=None,
help="Where to write temporary files (DIR)")
parser.add_argument("--debug", action="store_true")
args = parser.parse_args()
if args.asmdir.endswith("/"):
args.asmdir = args.asmdir[:-1]
if args.tempDir is None:
args.tempDir = args.asmdir
setupLogging(args.debug)
return args
def parseArgs():
parser = argparse.ArgumentParser(description=USAGE, \
formatter_class=argparse.RawDescriptionHelpFormatter)
parser.add_argument("reads", metavar="reads", type=str, \
help="Input reads .fasta or .fastq")
parser.add_argument("-t", "--target", type=str, \
help="Target sequence name")
parser.add_argument("-T", "--Target", type=str, \
help="Fasta file containing target sequence")
parser.add_argument("-s", "--super", dest="super", action="store_true",\
help="Treat each read as the target once")
parser.add_argument("-m", "--maxtail", type = int, default=sys.maxint, \
help="Max number of bases allowed to be in tail (inf)")
parser.add_argument("-n", "--nproc", dest="nproc", default=1, type=int,\
help="Number of processors to use with blasr (1)")
parser.add_argument("-o", "--outname", dest="outname", default="polish.out", \
type=str, \
help="Base name for output files (polish.out)")
parser.add_argument("--debug", action="store_true")
args = parser.parse_args()
setupLogging(args.debug)
#I don't think this is exhaustive
if (args.target is not None and args.Target is not None) \
or (args.super and (args.target is not None or args.Target is not None)):
print "Error! only specify one of --super or --target or --Target"
exit(1)
return args
def __init__(self):
"""
Given a protocol fn, load it up so we are ready to run.
"""
self.parseArgs()
setupLogging(self.options.debug)
sys.stderr.write("""
Please Cite: English, Adam C., Stephen Richards, Yi Han, Min Wang,
Vanesa Vee, Jiaxin Qu, Xiang Qin, et al. "Mind the
Gap: Upgrading Genomes with Pacific Biosciences RS
Long-Read Sequencing Technology." PLoS ONE 7, no. 11
(November 21, 2012): e47768.
doi:10.1371/journal.pone.0047768.\n\n""")
self.parseProtocol()
def parseArgs(argv):
parser = argparse.ArgumentParser(prog="Honey.py pie", description=USAGE, \
formatter_class=argparse.RawDescriptionHelpFormatter)
parser.add_argument("input", metavar="[SAM,BAM,FASTA,FASTQ,FOFN]", type=str, \
help="Input reads to be mapped")
parser.add_argument("reference", metavar="REFERENCE", type=str,\
help="Reference to map tails")
parser.add_argument("-o", "--output", type=str, default=None, \
help="Output Name (BAM.tails.[sam|bam])")
parser.add_argument("-t", "--minTail", type=int, default=100,\
help="Minimum tail length to attempt remapping (%(default)s)")
parser.add_argument("-n", "--nproc", type=int, default=1,\
help="Number of processors to use (%(default)s)")
parser.add_argument("-p", "--params", type=str, default=BLASRPARAMS,\
help="Specify custom blasr params. use -p=\"string\"")
parser.add_argument("--temp",
type=str,
default=tempfile.gettempdir(),
help="Where to save temporary files")
parser.add_argument("--chunks", type=int, default=0, \
help=("Create N scripts containing commands to "
"each input of the fofn (%(default)s)"))
parser.add_argument("--debug", action="store_true")
args = parser.parse_args(argv)
setupLogging(args.debug)
checkBlasrParams(args.params)
if args.output is None:
ext = args.input[args.input.rindex('.'):]
main = args.input[:args.input.rindex('.')]
if ext in [".sam", ".bam"]:
args.output = main + ".tails" + ext
else:
args.output = main + ".tails.sam"
return args
def parseArgs(argv):
parser = argparse.ArgumentParser(prog="Honey.py pie", description=USAGE, \
formatter_class=argparse.RawDescriptionHelpFormatter)
parser.add_argument("input", metavar="[SAM,BAM,FASTA,FASTQ,FOFN]", type=str, \
help="Input reads to be mapped")
parser.add_argument("reference", metavar="REFERENCE", type=str,\
help="Reference to map tails")
parser.add_argument("-o", "--output", type=str, default=None, \
help="Output Name (BAM.tails.[sam|bam])")
parser.add_argument("-t", "--minTail", type=int, default=100,\
help="Minimum tail length to attempt remapping (%(default)s)")
parser.add_argument("-n", "--nproc", type=int, default=1,\
help="Number of processors to use (%(default)s)")
parser.add_argument("-p", "--params", type=str, default=BLASRPARAMS,\
help="Specify custom blasr params. use -p=\"string\"")
parser.add_argument("--temp", type=str, default=tempfile.gettempdir(),
help="Where to save temporary files")
parser.add_argument("--chunks", type=int, default=0, \
help=("Create N scripts containing commands to "
"each input of the fofn (%(default)s)"))
parser.add_argument("--debug", action="store_true")
args = parser.parse_args(argv)
setupLogging(args.debug)
checkBlasrParams(args.params)
if args.output is None:
ext = args.input[args.input.rindex('.'):]
main = args.input[:args.input.rindex('.')]
if ext in [".sam", ".bam"]:
args.output = main + ".tails" + ext
else:
args.output = main + ".tails.sam"
return args
def parseArgs(argv):
parser = argparse.ArgumentParser(prog="Honey.py tails", description=USAGE, \
formatter_class=argparse.RawDescriptionHelpFormatter)
parser.add_argument("bam", metavar="BAM", type=str, \
help="BAM containing mapped reads")
parser.add_argument("-B", "--buffer", type=int, default=1000, \
help=("Buffer around breaks reads must fall "
"within to become clustered (%(default)s)"))
parser.add_argument("-b", "--minBreads", type=int, default=3,\
help="Minimum number of reads (%(default)s)")
parser.add_argument("-z", "--minZMWs", type=int, default=3, \
help="Minimum number of unique ZMWs (%(default)s)")
parser.add_argument("-q", "--minMapq", type=int, default=150, \
help="Minimum mapping quality of a read and it's tail to consider (%(default)s)")
parser.add_argument("-f", "--fastq", action="store_true", \
help="Write fastq for each cluster into a .tgz archive (%(default)s)")
parser.add_argument("-o", "--output", type=str, default=None, \
help="Output file to write results (BAM.hon.tails)")
# parser.add_argument("--noAdaptFilter", action="store_false", \
# help="Keep reads that appear to have a missed adapter orientation")
#parser.add_argument("-a", "--ambigous", action="store_true",
#help="Report SVs with ambigous annotation e.g. INS* (False)")
parser.add_argument("--debug", action="store_true")
parser.add_argument("-v", "--verboseFile", action="store_true", \
help="Print each read inside of a cluster to <output>.verbose (%(default)s)")
args = parser.parse_args(argv)
global BUFFER
BUFFER = args.buffer
if args.output is None:
args.output = args.bam[:-4] + ".hon.tails"
if args.verboseFile:
args.verboseFile = args.output + '.verbose'
setupLogging(args.debug)
return args
def parseArgs(argv):
parser = argparse.ArgumentParser(prog="Honey.py tails", description=USAGE, \
formatter_class=argparse.RawDescriptionHelpFormatter)
parser.add_argument("bam", metavar="BAM", type=str, \
help="BAM containing mapped reads")
parser.add_argument("-B", "--buffer", type=int, default=1000, \
help=("Buffer around breaks reads must fall "
"within to become clustered (%(default)s)"))
parser.add_argument("-b", "--minBreads", type=int, default=3,\
help="Minimum number of reads (%(default)s)")
parser.add_argument("-z", "--minZMWs", type=int, default=3, \
help="Minimum number of unique ZMWs (%(default)s)")
parser.add_argument("-q", "--minMapq", type=int, default=150, \
help="Minimum mapping quality of a read and it's tail to consider (%(default)s)")
parser.add_argument("-f", "--fastq", action="store_true", \
help="Write fastq for each cluster into a .tgz archive (%(default)s)")
parser.add_argument("-o", "--output", type=str, default=None, \
help="Output file to write results (BAM.hon.tails)")
# parser.add_argument("--noAdaptFilter", action="store_false", \
# help="Keep reads that appear to have a missed adapter orientation")
#parser.add_argument("-a", "--ambigous", action="store_true",
#help="Report SVs with ambigous annotation e.g. INS* (False)")
parser.add_argument("--debug", action="store_true")
parser.add_argument("-v", "--verboseFile", action="store_true", \
help="Print each read inside of a cluster to <output>.verbose (%(default)s)")
args = parser.parse_args(argv)
global BUFFER
BUFFER = args.buffer
if args.output is None:
args.output = args.bam[:-4] + ".hon.tails"
if args.verboseFile:
args.verboseFile = args.output + '.verbose'
setupLogging(args.debug)
return args
#!/usr/bin/python
import sys, logging, argparse
import pysam
from pbsuite.honey.HSpots import *
from pbsuite.utils.setupLogging import setupLogging
USAGE = "Recall spots in a hon.h5 file"
args = parseArgs(sys.argv[1:], established=True)
setupLogging(args.debug)
f = h5py.File(args.hon, 'a')
bam = pysam.Samfile(args.bam)
tsp = 0
makeKernals(args.binsize)
print "#CHROM\tOUTERSTART\tSTART\tINNERSTART\tINNEREND\tEND\tOUTEREND\tTYPE\tSIZE\tINFO"
for chrom in f.keys():
logging.info("Calling %s" % (chrom))
container = f[chrom]["data"]
start = f[chrom].attrs["start"]
spots = callHotSpots(container, start, args)
logging.info("Filtering spots")
fspot = 0
for spot in spots:
spot.chrom = chrom
spot.offset(start)
spot.estimateSize()
#the sv spans too far
if spot.size > args.spanMax:
#!/usr/bin/env python
import sys, logging, argparse
import pysam
from pbsuite.honey.HSpots import *
from pbsuite.utils.setupLogging import setupLogging
USAGE = "Recall spots in a hon.h5 file"
args = parseArgs(sys.argv[1:], established=True)
setupLogging(args.debug)
f = h5py.File(args.hon,'a')
bam = pysam.Samfile(args.bam)
tsp = 0
makeKernals(args.binsize)
print "#CHROM\tOUTERSTART\tSTART\tINNERSTART\tINNEREND\tEND\tOUTEREND\tTYPE\tSIZE\tINFO"
for chrom in f.keys():
logging.info("Calling %s" % (chrom))
container = f[chrom]["data"]
start = f[chrom].attrs["start"]
spots = callHotSpots(container, start, args)
logging.info("Filtering spots")
fspot = 0
for spot in spots:
spot.chrom = chrom
spot.offset(start)
spot.estimateSize()
#the sv spans too far
