def score(args):
"""
%prog score blastfile query.fasta A.ids
Add up the scores for each query seq. Go through the lines and for each
query sequence, add up the scores when subject is in each pile by A.ids.
"""
from jcvi.formats.base import SetFile
from jcvi.formats.fasta import Fasta
p = OptionParser(score.__doc__)
opts, args = p.parse_args(args)
if len(args) != 3:
sys.exit(not p.print_help())
blastfile, fastafile, idsfile = args
ids = SetFile(idsfile)
blast = Blast(blastfile)
scores = defaultdict(int)
for b in blast:
query = b.query
subject = b.subject
if subject not in ids:
continue
scores[query] += b.score
logging.debug("A total of {0} ids loaded.".format(len(ids)))
f = Fasta(fastafile)
for s in f.iterkeys_ordered():
sc = scores.get(s, 0)
print "\t".join((s, str(sc)))
def score(args):
"""
%prog score blastfile query.fasta A.ids
Add up the scores for each query seq. Go through the lines and for each
query sequence, add up the scores when subject is in each pile by A.ids.
"""
from jcvi.formats.base import SetFile
from jcvi.formats.fasta import Fasta
p = OptionParser(score.__doc__)
opts, args = p.parse_args(args)
if len(args) != 3:
sys.exit(not p.print_help())
blastfile, fastafile, idsfile = args
ids = SetFile(idsfile)
blast = Blast(blastfile)
scores = defaultdict(int)
for b in blast:
query = b.query
subject = b.subject
if subject not in ids:
continue
scores[query] += b.score
logging.debug("A total of {0} ids loaded.".format(len(ids)))
f = Fasta(fastafile)
for s in f.iterkeys_ordered():
sc = scores.get(s, 0)
print "\t".join((s, str(sc)))
def script(args):
"""
%prog script gffile cdna.fasta genome.fasta
Parse gmap gff and produce script for sim4db to refine.
"""
p = OptionParser(script.__doc__)
opts, args = p.parse_args(args)
if len(args) != 3:
sys.exit(p.print_help())
gffile, cdnafasta, genomefasta = args
scriptfile = gffile + ".script"
gff = Gff(gffile)
fw = open(scriptfile, "w")
cdnas = Fasta(cdnafasta, lazy=True)
cdnas = dict((x, i) for (i, x) in enumerate(cdnas.iterkeys_ordered()))
genomes = Fasta(genomefasta, lazy=True)
genomes = dict((x, i) for (i, x) in enumerate(genomes.iterkeys_ordered()))
extra = 50000 # 50-kb region surrounding the locus
for g in gff:
if g.type != "mRNA":
continue
cdna = g.attributes["Name"][0]
genome = g.seqid
ci = cdnas[cdna]
gi = genomes[genome]
strand = "-r" if g.strand == "-" else "-f"
start, end = g.start, g.end
start = max(0, start - extra)
end += extra
print >> fw, "{0} -e {1} -D {2} {3} {4}"\
.format(strand, ci, gi, start, end)
def script(args):
"""
%prog script gffile cdna.fasta genome.fasta
Parse gmap gff and produce script for sim4db to refine.
"""
p = OptionParser(script.__doc__)
opts, args = p.parse_args(args)
if len(args) != 3:
sys.exit(p.print_help())
gffile, cdnafasta, genomefasta = args
scriptfile = gffile + ".script"
gff = Gff(gffile)
fw = open(scriptfile, "w")
cdnas = Fasta(cdnafasta, lazy=True)
cdnas = dict((x, i) for (i, x) in enumerate(cdnas.iterkeys_ordered()))
genomes = Fasta(genomefasta, lazy=True)
genomes = dict((x, i) for (i, x) in enumerate(genomes.iterkeys_ordered()))
extra = 50000 # 50-kb region surrounding the locus
for g in gff:
if g.type != "mRNA":
continue
cdna = g.attributes["Name"][0]
genome = g.seqid
ci = cdnas[cdna]
gi = genomes[genome]
strand = "-r" if g.strand == "-" else "-f"
start, end = g.start, g.end
start = max(0, start - extra)
end += extra
print >> fw, "{0} -e {1} -D {2} {3} {4}"\
.format(strand, ci, gi, start, end)
def prepare(args):
"""
%prog prepare --rearray_lib=<rearraylibrary> --orig_lib_file=<origlibfile>
Inferred file names
---------------------------------------------
`lookuptblfile` : rearraylibrary.lookup
`rearraylibfile`: rearraylibrary.fasta
Pick sequences from the original library file and the rearrayed library file
based on the mapping information provided in the `lookuptblfile`.
# lookuptblfile format: column number (index)
# 1 (0) 2 (1) 3 (2) 4 (3) 5 (4) 6 (5)
# source_clone source_plate source_well dest_clone dest_plate dest_well
The 1st and 4th column in the `lookuptblfile` form the pair of clones which
constitute the elements used for the per-clone assembly.
"""
from operator import itemgetter
from jcvi.formats.fasta import Fasta, SeqIO
p = OptionParser(prepare.__doc__)
p.add_option("--rearray_lib",
default=None,
help="name of the rearrayed library [default: %default]")
p.add_option(
"--orig_lib_file",
help=
"fasta file containing reads from the original libraries [default: %default]"
)
g = OptionGroup(p, "Optional parameters")
g.add_option(
"--output_folder",
default="to_assemble",
help="output folder to write the FASTA files to [default: %default]")
p.add_option_group(g)
opts, args = p.parse_args(args)
if not opts.rearray_lib or not opts.orig_lib_file:
logging.error("Please specify the required parameters")
sys.exit(not p.print_help())
rearraylib, origlibfile = opts.rearray_lib, opts.orig_lib_file
if not op.isfile(origlibfile):
logging.error(
"Original library reads file `{0}` does not exist!".format(
origlibfile))
sys.exit()
lookuptblfile = rearraylib + '.lookup'
logging.debug(lookuptblfile)
if not op.isfile(lookuptblfile):
logging.error(
"Lookup table file `{0}` does not exist!".format(lookuptblfile))
sys.exit()
rearraylibfile = rearraylib + '.fasta'
logging.debug(rearraylibfile)
if not op.isfile(rearraylibfile):
logging.error(
"Rearrayed library reads file `{0}` does not exist!".format(
rearraylibfile))
sys.exit()
origlibFasta = Fasta(origlibfile)
rearraylibFasta = Fasta(rearraylibfile)
origlibids = [o for o in origlibFasta.iterkeys_ordered()]
rearraylibids = [r for r in rearraylibFasta.iterkeys_ordered()]
if not op.isdir(opts.output_folder):
logging.warning(
"Output directory `{0}` missing. Creating it now...".format(
opts.output_folder))
os.makedirs(opts.output_folder)
logfile = rearraylib + '.log'
log = open(logfile, 'w')
fp = open(lookuptblfile, 'r')
for row in fp:
origprefix, rearrayprefix = itemgetter(0, 3)(row.split('\t'))
libpair = origprefix + '_' + rearrayprefix
outfile = opts.output_folder + '/' + libpair + '.fasta'
ofp = open(outfile, 'w')
for o in origlibids:
if re.match(origprefix, o):
SeqIO.write(origlibFasta[o], ofp, 'fasta')
for r in rearraylibids:
if re.match(rearrayprefix, r):
SeqIO.write(rearraylibFasta[r], ofp, 'fasta')
ofp.close()
print(outfile, file=log)
log.close()
logging.debug('Wrote log file `{0}`'.format(logfile))
def prepare(args):
"""
%prog prepare --rearray_lib=<rearraylibrary> --orig_lib_file=<origlibfile>
Inferred file names
---------------------------------------------
`lookuptblfile` : rearraylibrary.lookup
`rearraylibfile`: rearraylibrary.fasta
Pick sequences from the original library file and the rearrayed library file
based on the mapping information provided in the `lookuptblfile`.
# lookuptblfile format: column number (index)
# 1 (0) 2 (1) 3 (2) 4 (3) 5 (4) 6 (5)
# source_clone source_plate source_well dest_clone dest_plate dest_well
The 1st and 4th column in the `lookuptblfile` form the pair of clones which
constitute the elements used for the per-clone assembly.
"""
from operator import itemgetter
from jcvi.formats.fasta import Fasta, SeqIO
p = OptionParser(prepare.__doc__)
p.add_option("--rearray_lib", default=None,
help="name of the rearrayed library [default: %default]")
p.add_option("--orig_lib_file",
help="fasta file containing reads from the original libraries [default: %default]")
g = OptionGroup(p, "Optional parameters")
g.add_option("--output_folder", default="to_assemble",
help="output folder to write the FASTA files to [default: %default]")
p.add_option_group(g)
opts, args = p.parse_args(args)
if not opts.rearray_lib or not opts.orig_lib_file:
logging.error("Please specify the required parameters")
sys.exit(not p.print_help())
rearraylib, origlibfile = opts.rearray_lib, opts.orig_lib_file
if not op.isfile(origlibfile):
logging.error("Original library reads file `{0}` does not exist!".format(origlibfile))
sys.exit()
lookuptblfile = rearraylib + '.lookup'
logging.debug(lookuptblfile)
if not op.isfile(lookuptblfile):
logging.error("Lookup table file `{0}` does not exist!".format(lookuptblfile))
sys.exit()
rearraylibfile = rearraylib + '.fasta'
logging.debug(rearraylibfile)
if not op.isfile(rearraylibfile):
logging.error("Rearrayed library reads file `{0}` does not exist!".format(rearraylibfile))
sys.exit()
origlibFasta = Fasta(origlibfile)
rearraylibFasta = Fasta(rearraylibfile)
origlibids = [o for o in origlibFasta.iterkeys_ordered()]
rearraylibids = [r for r in rearraylibFasta.iterkeys_ordered()]
if not op.isdir(opts.output_folder):
logging.warning("Output directory `{0}` missing. Creating it now...".format(opts.output_folder))
os.makedirs(opts.output_folder)
logfile = rearraylib + '.log'
log = open(logfile, 'w')
fp = open(lookuptblfile, 'r')
for row in fp:
origprefix, rearrayprefix = itemgetter(0,3)(row.split('\t'))
libpair = origprefix + '_' + rearrayprefix
outfile = opts.output_folder + '/' + libpair + '.fasta'
ofp = open(outfile, 'w')
for o in origlibids:
if re.match(origprefix, o):
SeqIO.write(origlibFasta[o], ofp, 'fasta')
for r in rearraylibids:
if re.match(rearrayprefix, r):
SeqIO.write(rearraylibFasta[r], ofp, 'fasta')
ofp.close()
print >>log, outfile
log.close()
logging.debug('Wrote log file `{0}`'.format(logfile))
