def screen(self, srrs=[]):
vrs_ctgs = {}
for i in srrs:
print("Screening {0}".format(i), file=sys.stderr)
s = screener.Screener(self.wd, i, self.dbs['virusdb'],
self.dbs['cdd'])
srr_alignments = s.screen_srr(s.srr, s.virus_db.path)
vdb_parser = s.vdbdump.run(s.srr, srr_alignments)
contigs = s.assemble(vdb_parser.dump_to_file())
putative_virus_contigs = s.cdd_screen(contigs, s.cdd_db.path,
os.path.join(s.wd, 'rpst'))
if len(putative_virus_contigs) > 0:
for j in putative_virus_contigs:
c = virus_contig.VirusContig(
"ctg_" + str(len(vrs_ctgs)),
s.assembler.parser.sequences[j].sequence, i,
s.assembler.parser.sequences[j].header, self.flank_len,
s.wd)
vrs_ctgs[c.name] = c
print("Prepared {} for budding".format(c.name))
print("Budding {} contigs".format(len(vrs_ctgs)))
s.bud(vrs_ctgs)
else:
print("No contigs with virus motifs detected")
sys.exit()
def screen(self, srrs=[]):
vrs_ctgs = {}
for i in srrs:
print("Screening {0}".format(i), file=sys.stderr)
s = screener.Screener(self.wd, i, self.dbs['virusdb'],
self.dbs['cdd'])
srr_alignments = s.screen_srr(s.srr, s.virus_db.path)
#vdb_parser = s.vdbdump.run(s.srr, srr_alignments)
sortit = s.sort_matches(self.sam, self.wd)
# use the argument as pattern
for index, filename in enumerate(args.in_files):
print('index = ', index, filename)
in_filename_phase.append(filename)
for index, filename in enumerate(args.out_files):
print('index = ', index, filename)
out_filename_phase.append(filename)
if debug_level > 1:
print('args :', len(sys.argv))
if debug_level > 1:
print('args :', len(sys.argv))
screener = screener.Screener()
screener.set_debug_level(debug_level)
if truncate_table:
screener.screener_table_reload(truncate_table)
screener.amfi_load_db()
screener.isin_load_db()
# screener.load_amfi_data(in_amfi_filename)
screener.screener_load_data(in_filename_phase[0])
screener.screener_dump_phase1(out_filename_phase[0])
screener.screener_dump_phase2(out_filename_phase[1])
screener.screener_dump_phase3(out_filename_phase[2], out_filename_phase[3],
def screen(self, inputs=[], intype="srr"):
vrs_ctgs = {}
blastcmds = []
if (intype == "srr"):
blastcmds.append("-srr")
elif (intype == "fasta"):
blastcmds.append("-query")
elif (intype == "fastq"):
blastcmds.append(["infmt", "fastq", "-query"])
else:
raise ValueError(
"Input type %s isn't supported; try srr, fasta or fastq, please."
% intype)
for i in inputs:
print("Screening {0}".format(i), file=sys.stderr)
s = screener.Screener(self.wd, i, self.dbs['virusdb'],
self.dbs['cdd'])
wd = os.path.join(self.wd, i)
### Need to figure out if the current input (i) is a file or an srr accession #.
intype = "srr"
if ('.' in i or not "SRR" in i[0:2]):
print("Input file detected: %s" % i)
intype = "query"
### Added logic here that checks for the existence of the sam file,
### and runs magicblast if it isn't there or is size 0
sambasename = "%s.sam" % i
samfile = os.path.join(wd, sambasename)
if (os.path.isfile(samfile) > 0):
print(
"There's already a sam file at %s, skipping magicblast." %
samfile)
else:
print(
"Could not find samfile at location %s, running magicblast."
% samfile)
s.screen_srr(s.srr, s.virus_db.path, samfile)
#srr_alignments = s.screen_srr(i, s.virus_db.path)
#vdb_parser = s.vdbdump.run(s.srr, srr_alignments)
samfile = os.path.join(wd, "magicblast.sam")
print("sam is %s and current working dir is %s" % (samfile, wd))
sortit = s.sort_matches(samfile, wd)
# vdb_parser = s.vdbdump.run(s.srr, srr_alignments)
# contigs = s.assemble(vdb_parser.dump_to_file())
weak_fasta = os.path.join(wd, "weak_%s.fasta" % i)
contigs = s.assemble(weak_fasta)
#print("contigs returns %s" % contigs)
sys.exit(0)
putative_virus_contigs = s.cdd_screen(contigs, s.cdd_db.path,
os.path.join(s.wd, 'rpst'))
if len(putative_virus_contigs) > 1000:
for j in putative_virus_contigs:
print("working on contig # j %s" %
s.assembler.parser.sequences)
c = virus_contig.VirusContig(
"ctg_" + str(len(vrs_ctgs)),
s.assembler.parser.sequences[j].sequence, i,
s.assembler.parser.sequences[j].header, self.flank_len,
s.wd)
vrs_ctgs[c.name] = c
print("Prepared {} for budding".format(c.name))
print("Budding {} contigs".format(len(vrs_ctgs)))
s.bud(vrs_ctgs)
else:
print("No contigs with virus motifs detected")
sys.exit()