def estimate(self, hits, outfile, placements):
hit = {}
logging.info("Estimating scores now")
if self.cfg["touch"]:
file.touch(outfile)
logging.info("Returning as we only touch")
return
r = base.readTSV(hits)
# count profile hits
for row in r:
if row["profile"] not in hit.keys():
hit[row["profile"]] = 1
else:
hit[row["profile"]] += 1
singletons = set(hit.keys())
multitons = set([k for k, v in hit.items() if v > 1])
# now we can estimate completeness and contamination for each placement
for i in range(len(placements)):
s = self.readSet(placements[i]["node"])
# completeness is the overap of both sets
cmpl = len(singletons & s) / len(s)
cont = len(multitons & s) / len(s)
# make to percentage and round to 2 positions
placements[i]["completeness"] = round(cmpl * 100, 2)
placements[i]["contamination"] = round(cont * 100, 2)
log("Finished estimating")
# write to output file
k = [
"completeness",
"contamination",
"node",
"n",
"ngenomes",
"cover",
"nPlacements",
"taxid",
"lineage",
"taxidlineage",
"file",
]
with open(outfile, "w") as f:
f.write("{}\n".format("\t".join(k)))
for p in placements:
# insert the file name
p["file"] = self.cfg["name"]
# write to file
f.write("{}\n".format("\t".join([str(p[key]) for key in k])))
log("Wrote estimates to: {}".format(outfile))
# done
return True
def get_silent_contig(self, fasta, hits, placements):
"""
given hits and the fasta file, we can calculate how many
DNA bp can not be detected as contaminant, because no markers
are on these contigs
"""
logging.debug("Calculating silent contig fraction")
if fasta is None:
logging.debug(
"As no DNA fasta was given, we can't compute the contig fraction"
)
for placement in placements:
placement["max_silent_contamination"] = "NA"
return
# get dict to link profile to contigs
profile_2_contigs = defaultdict(set)
for row in base.readTSV(hits):
profile_2_contigs[row["profile"]].add(row["chrom"])
# read in fasta once
dna_lens = {}
for rec in Fasta(fasta):
dna_lens[rec.name] = len(str(rec))
total_len = sum([v for k, v in dna_lens.items()])
# for each placement compute the silent fraction
for placement in placements:
contigs = set()
# make union of contigs
for profile in placement["set"]:
contigs = contigs | profile_2_contigs[profile]
# define missing contigs as contig names we did not locate any marker genes on
m_contigs = set(dna_lens.keys()) - contigs
missing_len = sum([dna_lens[contig] for contig in m_contigs])
placement["max_silent_contamination"] = round(
missing_len / total_len * 100, 2)
logging.debug("The silent fraction could be up to {}%".format(
placement["max_silent_contamination"]))
return
def estimate(self, hits, outfile, placements):
hit = {}
logging.info("Estimating scores now")
if self.cfg["touch"]:
file.touch(outfile)
logging.info("Returning as we only touch")
return
r = base.readTSV(hits)
# count profile hits
for row in r:
if row["profile"] not in hit.keys():
hit[row["profile"]] = 1
else:
hit[row["profile"]] += 1
singletons = set(hit.keys())
multitons = set([k for k, v in hit.items() if v > 1])
# now we can estimate completeness and contamination for each placement
for i in range(len(placements)):
s = self.readSet(placements[i]["node"])
placements[i]["set"] = s
# completeness is the overap of both sets
cmpl = len(singletons & s) / len(s)
cont = len(multitons & s) / len(s)
# make to percentage and round to 2 positions
placements[i]["completeness"] = round(cmpl * 100, 2)
placements[i]["contamination"] = round(cont * 100, 2)
# compute silent fraction per placement and set
self.get_silent_contig(self._clean_fasta, hits, placements)
log("Finished estimating")
self.write_outfile(outfile, placements)
# done
return True
def prepareAlignment(
self, pplaceAlinment, hmmerOutput, proteinList, proteinFasta, config, cfg, tmpDir,
):
# for each profile that we found a SCMG
# make concatenated fasta (including new sequence)
# align that to the hmm profile
profiles = []
with open(proteinList) as pl:
for l in pl:
profiles.append(l.strip())
# read in hmmer output to get seqnames of target proteins
# cols = []
# hmmer = []
# scmgsr = []
# with open(hmmerOutput) as ho:
# for line in ho:
# l = line.split()
# if len(cols) == 0:
# cols = l
# else:
# n = {}
# for k,v in zip(cols, l):
# n[k] = v
# hmmer.append(n)
# scmgsr.append(n['profile'])
# load hmmer results
hmmer = base.readTSV(hmmerOutput)
# extract SCMGs
scmgsr = [hit["profile"] for hit in hmmer]
# scmgs = [p for p in set(scmgsr) if scmgsr.count(p) == 1]
# db = [p for p in set(scmgsr) if scmgsr.count(p) == 2]
scmgs = scmgsr
# scmgs[-1] = db[0]
scmgs.sort()
# count the placements
self.lenscmgs = len(scmgs)
if self.lenscmgs > 0:
# get the protein names for each scmgs
proteinnames = []
# for all profiles extract proteins
hmmr = {}
# load into dict
for row in hmmer:
hmmr[row["profile"]] = row["subject"]
# make list of protein names
for m in scmgs:
proteinnames.append(hmmr[m])
# load proteins
proteins = Fasta(proteinFasta)
alignments = []
# for each protein/SCMG we need to make an alignment
for p, g in zip(scmgs, proteinnames):
# write seq to file
seq = proteins[g]
name = "{}_{}".format(p, g)
tmpfasta = os.path.join(tmpDir, "{}.faa".format(name))
geneAlignment = os.path.join(tmpDir, "{}.aln".format(name))
tmpfasta = base.writeFasta(tmpfasta, name, seq)
profileAlignment = config.pkgfile("{}.refpkg".format(p), "aln_fasta")
profileHMM = config.pkgfile("{}.refpkg".format(p), "profile")
# start and run alignment of this profile
ha = hmmalign("hmmalign", tmpfasta, geneAlignment)
ha.run(profileAlignment, profileHMM)
alignments.append(geneAlignment)
# after this we concatenate the alignments
# ordered by the profile name
self.input = base.horizontalConcat(
pplaceAlinment, alignments, scmgs, config.pkgfile("{}.refpkg".format("concat"), "aln_fasta"),
)