def search_one(self, query_seqid, pctid, threads=None):
pctid_str = "{:.1f}".format(pctid)
print("Searching", query_seqid, "at", pctid_str, "pct identity")
query_seq = self.seqs[query_seqid]
query_fp = "temp_query.fasta"
if os.path.exists(query_fp):
os.rename(query_fp, "temp_prev_query.fasta")
query_hits_fp = "temp_query_hits.txt"
if os.path.exists(query_hits_fp):
os.rename(query_hits_fp, "temp_prev_query_hits.txt")
with open(query_fp, "w") as f:
write_fasta(f, [(query_seqid, query_seq)])
aligner = PctidAligner(self.fasta_fp)
aligner.search(query_fp,
query_hits_fp,
min_pctid=pctid,
threads=threads,
max_hits=10000)
with open(query_hits_fp) as f:
hits = aligner.parse(f)
for hit in hits:
if hit["pident"] == pctid_str:
query = self.assemblies[hit["qseqid"]]
subject = self.assemblies[hit["sseqid"]]
pctid = hit["pident"]
yield AssemblyPair(query, subject, pctid, hit["qseqid"],
hit["sseqid"])
def search_reference_seqs(self, query_seqs):
query_file = tempfile.NamedTemporaryFile(suffix=".fasta", mode="wt")
write_fasta(query_file, query_seqs)
query_file.seek(0)
reference_hits_file = tempfile.NamedTemporaryFile(suffix=".txt",
mode="wt")
# 97.0 --> 0.97
vsearch_min_id = "{:.2f}".format(self.min_pct_id / 100)
vsearch_args = [
"vsearch",
"--usearch_global",
query_file.name,
"--db",
self.reference_udb_fp,
"--userout",
reference_hits_file.name,
"--iddef",
"2",
"--id",
vsearch_min_id,
"--maxaccepts",
self.max_hits,
"--userfields",
"query+target+id2+alnlen+mism+gaps+qilo+qihi+tilo+tihi+qs+ts+qrow+trow",
]
if self.num_threads:
vsearch_args.extend(["--threads", str(self.num_threads)])
subprocess.check_call(vsearch_args)
reference_hits_file.seek(0)
return reference_hits_file
def find_in_seqs(self, seqs):
if seqs.all_matched():
return
# Create the file paths
subject_fp = self._make_fp("subject_{0}.fa".format(self.suffix))
query_fp = self._make_fp("query_{0}.fa".format(self.suffix))
result_fp = self._make_fp("query_{0}.txt".format(self.suffix))
# Search
with open(subject_fp, "w") as f:
write_fasta(f, seqs.get_matched_offset0())
ba = VsearchAligner(subject_fp)
search_args = {
"min_id": round(self.min_pct_id / 100, 2),
"top_hits_only": None}
if self.cores > 0:
search_args["threads"] = self.cores
hits = ba.search(
seqs.get_unmatched_recs(), input_fp=query_fp, output_fp=result_fp,
**search_args)
# Refine
bext = HitExtender(seqs.get_unmatched_recs(), seqs.get_matched_offset0())
for hit in hits:
alignment = bext.extend_hit(hit)
subject_match = seqs.matches[alignment.subject_id]
aligned_region = AlignedRegion.from_subject(
alignment, subject_match.start, subject_match.end)
query_start_idx, query_end_idx = aligned_region.in_query()
query_offset = aligned_region.query_offset()
matchobj = PrimerMatch(
query_start_idx, query_end_idx, query_offset, "Alignment")
yield alignment.query_id, matchobj
def search(self, seqs, input_fp=None, output_fp=None, **kwargs):
if input_fp is None:
infile = tempfile.NamedTemporaryFile(mode="w+t", encoding="utf-8")
write_fasta(infile, seqs)
infile.seek(0)
input_fp = infile.name
else:
with open(input_fp, "w") as f:
write_fasta(f, seqs)
if output_fp is None:
outfile = tempfile.NamedTemporaryFile()
output_fp = outfile.name
self._call(input_fp, self.ref_seqs_fp, output_fp, **kwargs)
with open(output_fp) as f:
for hit in self._parse(f):
yield hit
def search_seq(self, query_seqid, query_seq, min_pctid=90.0, threads=None):
query_fp = "temp_query.fasta"
if os.path.exists(query_fp):
os.rename(query_fp, "temp_prev_query.fasta")
query_hits_fp = "temp_query_hits.txt"
if os.path.exists(query_hits_fp):
os.rename(query_hits_fp, "temp_prev_query_hits.txt")
with open(query_fp, "w") as f:
write_fasta(f, [(query_seqid, query_seq)])
aligner = PctidAligner(self.fasta_fp)
aligner.search(query_fp,
query_hits_fp,
min_pctid=min_pctid,
threads=threads,
max_hits=10000)
with open(query_hits_fp) as f:
hits = aligner.parse(f)
for hit in hits:
query = self.assemblies[hit["qseqid"]]
subject = self.assemblies[hit["sseqid"]]
pctid = hit["pident"]
if query.accession != subject.accession:
yield AssemblyPair(query, subject, pctid, hit["qseqid"],
hit["sseqid"])
def save(self):
with open(self.fasta_fp, "w") as f:
write_fasta(f, self.seqs.items())
with open(self.accession_fp, "w") as f:
for seqid, assembly in self.assemblies.items():
f.write("{0}\t{1}\n".format(seqid, assembly.accession))
def test_write_fasta(self):
f = tempfile.NamedTemporaryFile(mode="w+t", encoding="utf-8")
seqs = [("a", "CCGGT"), ("b", "TTTTTTTTT")]
write_fasta(f, seqs)
f.seek(0)
self.assertEqual(f.read(), ">a\nCCGGT\n>b\nTTTTTTTTT\n")