def dump_results(outfile, fastadir, contigs):
for r, d in contigs.iteritems():
with open(os.path.join(fastadir, "{}.fa".format(r)), "w") as outfasta:
for s,seq in d["seqs"].iteritems():
outfile.write( "{}:{}\t{}\t{}\n".format( "\t".join(str(d["region"]).strip().split()[0:5]), s, len(seq), "*" ) )
outfasta.write(">{}\n".format(s))
for km in kmers.kmerize(seq, 50, 50):
outfasta.write("{}\n".format(km))
ref += str(kmers[e].name)[kstep:]
## step through BWTs trying to assemble between flanking k-mers
for bb, b in enumerate(bwts):
alt, n = assembly.build_bridge(b, str(kmers[s-1].name), str(kmers[e].name))
# check for successful assembly
if not len(alt):
# unsuccessful assembly
logger.info("Couldn't bridge range {}:{}-{}".format(kmers[s-1].chrom, kmers[s-1].start+1, kmers[e].end))
else:
# assembly successful; align to reference
ref_aln, alt_aln, ops = swalign.align(ref, alt[0])
# check coverage of alt alleles
alt_counts = []
for k in ktools.kmerize(alt[0], ksize, kstep):
if len(k) == ksize:
alt_counts.append( util.count_reads(b, k) )
support = np.median(alt_counts)
#print(ref_aln)
#print(ops)
#print(alt_aln)
# iterate on variant sites in alignment
for offset, site_ref, site_alt in swalign.reconcile(ref_aln, alt_aln):
# generate VCF entry
CHROM = kmers[s-1].chrom
POS = kmers[s-1].start+offset+1 #NB: must convert to 1-based
ID = "."
REF = site_ref
ALT = site_alt
QUAL = "."
parser.add_argument( "-f","--fasta", type = argparse.FileType("rU"),
help = "fasta file containing sequences to search with" )
parser.add_argument( "-M", "--msbwt", type = io.readable_dir, nargs = "+",
help = "one or more msBWTs in which to count k-mers" )
parser.add_argument( "-k", "--kmer", type = int,
default = 0,
help = "k-mer size (set to 0 to use all of each sequence) [default: %(default)d]")
parser.add_argument( "--normalize", action = "store_true",
help = "normalize counts against total size of msBWTs [default: %(default)d]")
args = parser.parse_args()
bwts = util.BwtSet(args.msbwt)
fa = SeqIO.parse(args.fasta, format = "fasta")
sys.stderr.write( "Using the following msBWTs:\n{}".format(str(bwts)) )
sys.stderr.write( "Reporting counts as parts per billion.\n" )
if args.kmer > 0:
sys.stderr.write( "Breaking sequences into k-mers of length {} for searches.\n".format(args.kmer) )
else:
sys.stderr.write( "Searching with provided sequences as-is.\n" )
for seq in fa:
if args.kmer > 0:
k = args.kmer
else:
k = len(seq.seq)
for subseq in kmers.kmerize(seq.seq, k):
for bwtname,count in bwts.count(subseq, args.normalize).iteritems():
print seq.name, subseq, bwtname, count
all_found = True
these_contigs = collections.defaultdict(str)
for i in range(0, len(bwtnames)):
sys.stderr.write("\t{} ...\n".format(bwtnames[i]))
## first check that seed will work in this BWT
## allow step-down of seed size to get started, but hold k-mer size for assembly constant
seed_found = False
k_start = args.kmer + 1
while not seed_found:
k_start -= 1
## if seed is less than 21nt long, it's probably hopeless
if k_start < 21:
break
for seed_k in kmers.kmerize(seed, k_start, 1):
x = util.count_reads(msbwts[i], seed_k)
if x >= args.minweight:
seed_found = True
seed = seed_k
sys.stderr.write("\t\tseed accepted ({} bp): {}\n".format(len(seed_k), seed_k))
break
## seed not found in this sample: break loop
if not seed_found:
all_found = False
break
## do assembly
seq = assembly.greedy_assemble( args.msbwt[i], seed = seed, k = args.kmer, count_k = args.count_kmer, direction = True,
min_weight = args.minweight, max_weight = args.maxweight,
break_kmers.append( (kmer_stash[i_start].name, kmer_stash[i_end].name) )
break_counts.append( (kmer_stash[i_start].score, kmer_stash[i_end].score) )
for i in range(0, len(break_coords)):
hap = assemble_inward(msbwt[0], break_kmers[i][0], break_kmers[i][1])
if args.verbose:
outline = [ break_coords[i][0], break_coords[i][1], break_counts[i][0], break_coords[i][2], break_counts[i][1] ]
sys.stderr.write("\t".join([ str(x) for x in outline ]) + "\n")
sys.stderr.write("\t" + str(hap) + "\n\n")
for h in hap:
hap_avg = {}
for (seq, count) in h.iteritems():
if count > args.maf:
ungapped_seq = dna.ungap(seq)
k_counts = []
for k in kmers.kmerize(dna.ungap(seq), kmer_size):
if dna.complexity(k) > args.complexity:
k_counts.append( util.count_reads(msbwt[0], k) )
if len(k_counts):
hap_avg.update({ ungapped_seq: max(k_counts) })
flag = "*"
if len(hap_avg.keys()) > 1:
sys.stderr.write("Warning: apparently there is >1 haplotype at this variant site.\n")
flag = "+"
for (hh, hc) in hap_avg.iteritems():
if hc > args.maxhits:
print break_coords[i][0], break_coords[i][1], break_coords[i][2], hh, hc, flag
sys.stdout.flush() # force write; this lets me peek at output in almost-real time
if args.verbose:
for i in range(0, len(break_coords)):
hap = assemble_inward(msbwt[0], break_kmers[i][0], break_kmers[i][1])
if args.verbose:
outline = [
break_coords[i][0], break_coords[i][1], break_counts[i][0],
break_coords[i][2], break_counts[i][1]
]
sys.stderr.write("\t".join([str(x) for x in outline]) + "\n")
sys.stderr.write("\t" + str(hap) + "\n\n")
for h in hap:
hap_avg = {}
for (seq, count) in h.iteritems():
if count > args.maf:
ungapped_seq = dna.ungap(seq)
k_counts = []
for k in kmers.kmerize(dna.ungap(seq), kmer_size):
if dna.complexity(k) > args.complexity:
k_counts.append(util.count_reads(msbwt[0], k))
if len(k_counts):
hap_avg.update({ungapped_seq: max(k_counts)})
flag = "*"
if len(hap_avg.keys()) > 1:
sys.stderr.write(
"Warning: apparently there is >1 haplotype at this variant site.\n"
)
flag = "+"
for (hh, hc) in hap_avg.iteritems():
if hc > args.maxhits:
print break_coords[i][0], break_coords[i][1], break_coords[i][
2], hh, hc, flag
save_graph = True
bwtname = re.sub(r"/+$", "", args.msbwt).split("/").pop()
## make sure seed sequences have proper form
seed = dna.ungap(args.seed.upper())
sys.stderr.write("Seed sequence is: {}\n".format(seed))
end_seeds = []
if args.end_seeds:
for s in args.end_seeds:
end_seeds.append(dna.ungap(s))
## examine the seed, taking the first k-mer which returns a result
msbwts = util.load_bwts([args.msbwt])
seed_found = False
for seed_k in kmers.kmerize(seed, args.kmer, 1):
x = util.count_reads(msbwts[0], seed_k)
if x >= args.minweight:
seed_found = True
seed = seed_k
print seed_k
break
if not seed_found:
sys.exit(
"Seed sequence doesn't contain any k-mers which meet the specified abundance threshold."
)
## do assembly
seq = assembly.greedy_assemble(args.msbwt,
seed=seed,
save_graph = True
bwtname = re.sub(r"/+$","", args.msbwt).split("/").pop()
## make sure seed sequences have proper form
seed = dna.ungap(args.seed.upper())
sys.stderr.write("Seed sequence is: {}\n".format(seed))
end_seeds = []
if args.end_seeds:
for s in args.end_seeds:
end_seeds.append( dna.ungap(s) )
## examine the seed, taking the first k-mer which returns a result
msbwts = util.load_bwts([args.msbwt])
seed_found = False
for seed_k in kmers.kmerize(seed, args.kmer, 1):
x = util.count_reads(msbwts[0], seed_k)
if x >= args.minweight:
seed_found = True
seed = seed_k
print seed_k
break
if not seed_found:
sys.exit("Seed sequence doesn't contain any k-mers which meet the specified abundance threshold.")
## do assembly
seq = assembly.greedy_assemble( args.msbwt, seed = seed, end_seeds = end_seeds, k = args.kmer, count_k = args.count_kmer, direction = args.backward,
min_weight = args.minweight, max_weight = args.maxweight,
max_nodes = args.maxnodes, max_length = args.maxlength,
save = save_graph, outprefix = args.prefix, memmap = args.memmap )
for bb, b in enumerate(bwts):
alt, n = assembly.build_bridge(b, str(kmers[s - 1].name),
str(kmers[e].name))
# check for successful assembly
if not len(alt):
# unsuccessful assembly
logger.info("Couldn't bridge range {}:{}-{}".format(
kmers[s - 1].chrom, kmers[s - 1].start + 1,
kmers[e].end))
else:
# assembly successful; align to reference
ref_aln, alt_aln, ops = swalign.align(ref, alt[0])
# check coverage of alt alleles
alt_counts = []
for k in ktools.kmerize(alt[0], ksize, kstep):
if len(k) == ksize:
alt_counts.append(util.count_reads(b, k))
support = np.median(alt_counts)
#print(ref_aln)
#print(ops)
#print(alt_aln)
# iterate on variant sites in alignment
for offset, site_ref, site_alt in swalign.reconcile(
ref_aln, alt_aln):
# generate VCF entry
CHROM = kmers[s - 1].chrom
POS = kmers[
s -
1].start + offset + 1 #NB: must convert to 1-based
ID = "."