def main():
readoids = options.output + ".readoids.fasta"
rc = RunCommand([
'create_readoids.py', '-l',
str(options.rd_ln), '-d',
str(options.distance), '-w',
str(options.window_size), '-o', readoids, args[1]
])
sys.stderr.write("Running: " + rc.get_command())
rc.run_command()
detail_output = options.output + ".scaffold_accuracy"
rc = RunCommand(['run_nucmer.py', '-p', detail_output, args[0], readoids])
sys.stderr.write("Running: " + rc.get_command())
rc.run_command()
coords_file = detail_output + ".coords"
table_output = options.table + ".scaffold_accuracy"
rc = RunCommand([
'assess_readoids_coords.py', '-c', coords_file, '-p', readoids, '-f',
args[1], '-r', args[0], '-v', detail_output, '-l',
str(options.rd_ln), '-d',
str(options.distance), '-e',
str(options.error), '-o', table_output
])
sys.stderr.write("Running: " + rc.get_command())
rc.run_command()
return 0
def main():
# set up command
cmd = None
task = options.task
if options.make_blastdb:
type = 'nucl'
if options.is_protein:
type = 'prot'
cmd_list = [constant.MAKEBLASTDB, '-in', args[0], '-dbtype', type]
rc = RunCommand(cmd_list)
print "Running command: " + rc.get_command() + '\n'
rc.run_command()
if options.vec_screen:
cmd = NcbiblastnCommandline(query=args[1], db=args[0], evalue=700, outfmt=options.outfmt, reward=1,
penalty=-5, gapopen=3, gapextend=3, dust='yes', searchsp=1750000000000,
out=options.output, task=task, num_threads=options.threads)
elif options.ncbi_screen:
# Note: new NCBI requirements say to use lcase_masking (See GAAG-510 for documentation)
cmd = NcbiblastnCommandline(query=args[1], db=args[0], outfmt=options.outfmt, dust='yes',
perc_identity=90, lcase_masking='true', task='megablast',
out=options.output, num_threads=options.threads)
#Blast documentation says soft_masking option if not specified defaults to true so even though
#this biopython class does not support it the feature should be enabled by default
elif options.mito_screen:
# Note: new NCBI requirements say to use lcase_masking (See GAAG-510 for documentation)
cmd = NcbiblastnCommandline(query=args[1], db=args[0], outfmt=options.outfmt, dust='yes',
perc_identity=98.6, lcase_masking='true', task='blastn',
out=options.output, num_threads=options.threads)
#Blast documentation says soft_masking option if not specified defaults to true so even though
#this biopython class does not support it the feature should be enabled by default
elif options.rRNA_screen:
cmd = NcbiblastnCommandline(query=args[1], db=args[0], outfmt=options.outfmt, dust='yes',
perc_identity=95, lcase_masking='', task='megablast',
out=options.output, num_threads=options.threads, evalue=1e-9, window_size=120, gapextend=2, gapopen=4,
no_greedy='', penalty=-4, reward=3, word_size=12, xdrop_gap=20)
#Options not supported: in_pssm='',soft_masking='true', matrix=5000000, max_intron_length=18, db_gencode=3,
else:
if options.task in TASKS:
program = TASKS[options.task]
if re.search("Ncbiblastn", str(program)):
cmd = program(query=args[1], db=args[0], evalue=options.evalue,
outfmt=options.outfmt, out=options.output,
num_threads=options.threads, max_target_seqs=options.max_targets,
task=task)
else:
cmd = program(query=args[1], db=args[0], evalue=options.evalue,
outfmt=options.outfmt, out=options.output,
num_threads=options.threads, max_target_seqs=options.max_targets)
else:
print "Unrecognized blast task, " + options.task
sys.exit(-1)
print "Running BLAST command: " + str(cmd) + '\n'
out, err = cmd()
# print "out: out\terr: err"
return 0
def main():
if __check_inputs(options.classify, options.rdp_out):
rnammer_cmd = __build_rnammer_cmd(args[0], options.rnammer_out,
options.gene, options.superkingdom)
rc = RunCommand(rnammer_cmd)
print "RUNNING: " + rc.get_command()
rc.run_command()
if options.classify:
rdp_cmd = __build_rdp_cmd(options.rnammer_out, options.rdp_out)
rc = RunCommand(rdp_cmd)
print "RUNNING: " + rc.get_command()
rc.run_command()
return 0
else:
print "If classifying hits, must supply classify flag and rdp output file."
sys.exit(-1)
def main():
delta = args[0]
# get information about our inputs
reference, query = _get_query_and_reference_from_delta_file(delta)
query_seqs = _get_query_sequences(query)
reference_list, reference_lengths = _get_reference_details(reference)
# get alignment information
command = _get_show_tilings_command(options.output, options.id,
options.coverage, options.circular,
delta)
reference_tilings = _get_tilings_information(command)
# see if we need circular query information
circular_queries = {}
if options.circular:
command = _get_circular_show_tilings_command(options.output, delta)
circular_queries = _get_circular_alignments(command)
print "Ordering and orienting using", delta
print "Reference", reference
print "Query", query
# get ono information from our gathered data
ono_sequences_list, ono_details_list = _parse_tilings(
reference_list, reference_lengths, reference_tilings, query_seqs,
circular_queries)
# print out our output
_write_details_to_file(ono_details_list,
options.output + ".ono.details.txt")
interim_fasta = options.output + ".interim.fasta"
SeqIO.write(ono_sequences_list, interim_fasta, "fasta")
make_assembly_command = [
"make_standard_assembly_files.py", "-S", interim_fasta, "-o",
options.output + ".ono"
]
if options.rename:
make_assembly_command += ['-r']
rc = RunCommand(make_assembly_command)
print "Executing", rc.get_command()
rc.run_command()
return 0
def main():
delta = args[0]
delta_file = open(delta, 'r')
reference, query = delta_file.readline().rstrip().split(" ")
query_fasta = SimpleFastaFile(query)
query_seqs = query_fasta.get_sequence()
if os.path.isfile(options.output + ".interim.fasta"):
remove_command = "rm", options.output + ".interim.fasta"
rc = RunCommand(remove_command)
print rc.get_command()
rc.run_command()
ono_fasta = SimpleFastaFile(options.output + ".interim.fasta")
print "Ordering and orienting using", delta
print "Reference", reference
print "Query", query
arg_list = ['-R', '-v 5', '-V 0', '-u', options.output + '.unplaced']
if options.circular:
arg_list += ["-c"]
rc = RunCommand(__build_showtiling_command(delta, arg_list))
print "Executing:", rc.get_command()
print ""
left_cut = 0
for line in rc.run_command().splitlines():
wraps = False
if (line.startswith(">")):
#IF THERE WAS A WRAP ON THE PREVIOUS REFERENCE SEQUENCE
if (left_cut):
print left_text
ono_fasta.addSeq(left_query + "_left", left_side)
ono_fasta.reverseSeq(left_query + "_left")
query_seqs[left_query] = 0
ref, bases, null = line.rstrip().split(" ")
print "Ordering to", ref
left_cut = 0
else:
rstart, rend, qstart, qend, cov, id, ori, query = line.rstrip(
).split("\t")
#CHECK TO SEE IF THIS IS AN ALIGNMENT WHICH WRAPS AROUND THE END OF A REFERENCE SEQUENCE
if (int(rstart) < 0 and options.circular):
#print "WRAP", line.rstrip().split("\t")
wraps = True
arg_list = [
'-R', '-a', '-v 5', '-g -1', '-V 0', '-u',
options.output + '.unplaced', '-c'
]
rc = RunCommand(__build_showtiling_command(delta, arg_list))
#print "Executing:",rc.get_command()
for line in (rc.run_command().splitlines()):
align = line.rstrip().split("\t")
#print align
#print align[0], align[12], query
if (int(align[0]) == 1 and align[12] == query):
#print "CLEAN WRAP FOUND!", align
#PRINT RIGHT SIDE OF QUERY AND STORE LEFT CUT SITE
if (ori == "+"):
print "\tStoring left side of the query to print at the end of alignments to this reference sequence"
left_cut = int(align[2]) - 1
left_side = query_seqs[query][:left_cut]
left_query = query
left_text = "\tWriting left side of " + query + " (1 to " + str(
left_cut
) + ") aligned " + cov + " at " + id + "% identity in the forward orientation."
right_cut = int(align[2]) - 1
print "\tWriting right side of " + query + " (" + str(
right_cut + 1
) + " to end) aligned " + cov + " at " + id + "% identity starting at 1 in the forward orientation."
right_side = query_seqs[query][right_cut:]
ono_fasta.addSeq(query + "_right", right_side)
else:
print "\tStoring left side of the query to print at the end of alignments to this reference sequence"
left_cut = int(align[2])
left_side = query_seqs[query][left_cut:]
left_query = query
left_text = "\tWriting left side of " + query + "(" + str(
left_cut + 1
) + " to end) aligned " + cov + " at " + id + "% identity in the reverse orientation."
right_cut = int(align[2])
print "\tWriting right side of " + query + " (1 to " + str(
right_cut
) + ") aligned " + cov, "at", id + "% identity starting at 1 in the reverse orientation."
right_side = query_seqs[query][:right_cut]
ono_fasta.addSeq(query + "_right", right_side)
ono_fasta.reverseSeq(query + "_right")
if not left_cut:
print "\tIt appears this query sequence overlaps, but no clean cut site was found. Printing as normal."
wraps = False
#IF THERE IS NO WRAP
if (not wraps):
if ori == "+":
print "\tWriting", query, "aligned", cov, "at", id + "% identity starting at", rstart, "in the forward orientation."
ono_fasta.addSeq(query, query_seqs[query])
else:
print "\tWriting", query, "aligned", cov, "at", id + "% identity starting at", rstart, "in the reverse orientation."
ono_fasta.addSeq(query, query_seqs[query])
ono_fasta.reverseSeq(query)
query_seqs[query] = 0
#IF THE LAST REFERENCE HAD A WRAP, PRINT LEFT SIDE
if (left_cut):
print left_text
ono_fasta.addSeq(left_query + "_left", left_side)
if "reverse" in left_text:
ono_fasta.reverseSeq(left_query + "_left")
query_seqs[left_query] = 0
#PRINT ANY UNALIGNED SEQUENCES
for contig in query_seqs:
if query_seqs[contig]:
print "Writing unaligned sequence:", contig
ono_fasta.addSeq(contig, query_seqs[contig])
ono_fasta.writeSeqFile()
make_assembly_command = [
"make_standard_assembly_files.py", "-S",
options.output + ".interim.fasta", "-o", options.output + ".ono"
]
if options.rename:
make_assembly_command += ['-r']
rc = RunCommand(make_assembly_command)
print "Executing", rc.get_command()
rc.run_command()
def main():
# set up command
cmd = None
task = options.task
if options.make_blastdb:
type = 'nucl'
if options.is_protein:
type = 'prot'
cmd_list = [constant.MAKEBLASTDB, '-in', args[0], '-dbtype', type]
rc = RunCommand(cmd_list)
print "Running command: " + rc.get_command() + '\n'
rc.run_command()
if options.vec_screen:
cmd = NcbiblastnCommandline(query=args[1],
db=args[0],
evalue=700,
outfmt=options.outfmt,
reward=1,
penalty=-5,
gapopen=3,
gapextend=3,
dust='yes',
searchsp=1750000000000,
out=options.output,
task=task,
num_threads=options.threads)
elif options.ncbi_screen:
cmd = NcbiblastnCommandline(query=args[1],
db=args[0],
outfmt=options.outfmt,
dust='yes',
perc_identity=90,
lcase_masking='',
task='megablast',
out=options.output,
num_threads=options.threads)
#soft_masking='true',
elif options.rRNA_screen:
cmd = NcbiblastnCommandline(query=args[1],
db=args[0],
outfmt=options.outfmt,
dust='yes',
perc_identity=95,
lcase_masking='',
task='megablast',
out=options.output,
num_threads=options.threads,
evalue=1e-9,
window_size=120,
gapextend=2,
gapopen=4,
no_greedy='',
penalty=-4,
reward=3,
word_size=12,
xdrop_gap=20)
#Options not supported: in_pssm='',soft_masking='true', matrix=5000000, max_intron_length=18, db_gencode=3,
else:
if options.task in TASKS:
program = TASKS[options.task]
if re.search("Ncbiblastn", str(program)):
cmd = program(query=args[1],
db=args[0],
evalue=options.evalue,
outfmt=options.outfmt,
out=options.output,
num_threads=options.threads,
max_target_seqs=options.max_targets,
task=task)
else:
cmd = program(query=args[1],
db=args[0],
evalue=options.evalue,
outfmt=options.outfmt,
out=options.output,
num_threads=options.threads,
max_target_seqs=options.max_targets)
else:
print "Unrecognized blast task, " + options.task
sys.exit(-1)
print "Running BLAST command: " + str(cmd) + '\n'
out, err = cmd()
return 0
