def analyse(args):
drug_loci = pp.load_bed(args.bed, [6], 4)
for d in drug_loci:
drug_loci[d] = [x.lower() for x in drug_loci[d][0].split(";")]
print(drug_loci)
drug = args.drug
results = json.load(open("results.json"))
fn = defaultdict(int)
fp = defaultdict(int)
print("Number true positives: %s" % len(results[drug]["tp"]))
print("Number false negatives: %s" % len(results[drug]["fn"]))
print("Number true negatives: %s" % len(results[drug]["tn"]))
print("Number false positives: %s" % len(results[drug]["fp"]))
for s in results[drug]["fn"]:
print("%s\tFalse_negative" % s)
tmp = json.load(open("%s/%s.results.json" % (args.dir, s)))
for var in tmp["other_variants"]:
if not any([drug in x for x in drug_loci[var["locus_tag"]]]):
continue
fn[(var["gene"], var["locus_tag"], var["change"])] += 1
for s in results[drug]["fp"]:
print("%s\tFalse_positive" % s)
tmp = json.load(open("%s/%s.results.json" % (args.dir, s)))
for var in tmp["dr_variants"]:
#if drug not in drug_loci[var["locus_tag"]]:
if drug != var["drug"].lower():
continue
fp[(var["gene"], var["locus_tag"], var["change"])] += 1
for key in sorted(fn, key=lambda x: fn[x]):
gene, rv, var = key
if gene == ".":
gene = rv
print("False_Negative\t%s\t%s\t%s" % (gene, var, fn[key]))
for key in sorted(fp, key=lambda x: fp[x]):
gene, rv, var = key
# tmp = odds_ratio[rv][var]
# print "False_Positive\t%s\t%s\t%s\t%s\t%s\t%s" % (gene,var,fp[key],tmp[4],tmp[5],tmp[6])
print("False_Positive\t%s\t%s\t%s" % (gene, var, fp[key]))
if args.itol:
write_itol(results[drug]["fp"], results[drug]["fn"], args.drug)
def calculate(args):
sample_file = args.samples
dst_file = args.dst
dst = load_dst(dst_file)
drug_loci = pp.load_bed(args.bed, [6], 4) # {'Rv0668': ('rifampicin')}
FAIL = open("samples_not_found.txt", "w")
samples = [x.rstrip() for x in open(sample_file).readlines()]
ext = ".results.json"
drugs = [d.lower() for d in dst[samples[0]].keys()]
results = {
d: {
"tp": [],
"tn": [],
"fp": [],
"fn": []
}
for d in drugs + ["flq", "mdr", "xdr", "sus"]
}
counts = {
d: {
"tp": 0,
"tn": 0,
"fp": 0,
"fn": 0
}
for d in drugs + ["flq", "mdr", "xdr", "sus"]
}
pre = args.dir if args.dir else ""
for s in tqdm(samples):
res_file = "%s/%s%s" % (pre, s, ext)
if pp.nofile(res_file):
pp.log("Warning: %s does not exist!" % res_file)
FAIL.write("%s\n" % s)
continue
res = json.load(open(res_file))
na_drugs = set()
for locus in drug_loci:
if res["missing_regions"][locus] > args.miss:
for tmp in drug_loci[locus][0].split(","):
na_drugs.add(tmp)
resistant_drugs = [d["drug"].lower() for d in res["dr_variants"]]
for d in drugs:
if d in na_drugs:
dst[s][d] = "NA"
for d in drugs:
if dst[s][d] == "0" and d not in resistant_drugs:
results[d]["tn"].append(s)
counts[d]["tn"] += 1
elif dst[s][d] == "0" and d in resistant_drugs:
results[d]["fp"].append(s)
counts[d]["fp"] += 1
elif dst[s][d] == "1" and d not in resistant_drugs:
results[d]["fn"].append(s)
counts[d]["fn"] += 1
elif dst[s][d] == "1" and d in resistant_drugs:
results[d]["tp"].append(s)
counts[d]["tp"] += 1
#### Fluoroquinolones ####
dst_flq = "0"
dst_flq_NA = True
for d in fluoroquinolones:
if d not in dst[s]: continue
if dst[s][d] != "NA": dst_flq_NA = False
if dst[s][d] == "1": dst_flq = "1"
dst_flq_list = [dst[s][d] for d in fluoroquinolones if d in dst[s]]
if "1" in dst_flq_list and "0" in dst_flq_list:
dst_flq = "NA"
if dst_flq_NA: dst_flq = "NA"
gst_flq = "0"
for d in fluoroquinolones:
if d in resistant_drugs: gst_flq = "1"
if dst_flq == "1" and gst_flq == "1":
results["flq"]["tp"].append(s)
counts["flq"]["tp"] += 1
if dst_flq == "0" and gst_flq == "1":
results["flq"]["fp"].append(s)
counts["flq"]["fp"] += 1
if dst_flq == "1" and gst_flq == "0":
results["flq"]["fn"].append(s)
counts["flq"]["fn"] += 1
if dst_flq == "0" and gst_flq == "0":
results["flq"]["tn"].append(s)
counts["flq"]["tn"] += 1
#### MDR & XDR ####
dst_mdr = "1" if dst[s]["rifampicin"] == "1" and dst[s][
"isoniazid"] == "1" else "0"
if dst[s]["rifampicin"] == "NA" or dst[s]["isoniazid"] == "NA":
dst_mdr = "NA"
flq = False
flq_NA = True
for d in fluoroquinolones:
if d not in dst[s]: continue
if dst[s][d] != "NA": flq_NA = False
if dst[s][d] == "1": flq = True
amg = False
amg_NA = True
for d in aminoglycosides:
if d not in dst[s]: continue
if dst[s][d] != "NA": amg_NA = False
if dst[s][d] == "1": amg = True
dst_xdr = "1" if dst_mdr == "1" and flq and amg else "0"
if flq_NA or amg_NA: dst_xdr = "NA"
if dst_mdr == "NA": dst_xdr = "NA"
#### Profiling results #####
gst_mdr = "1" if "rifampicin" in resistant_drugs and "isoniazid" in resistant_drugs else "0"
flq = False
for d in fluoroquinolones:
if d in resistant_drugs: flq = True
amg = False
for d in aminoglycosides:
if d in resistant_drugs: amg = True
gst_xdr = "1" if gst_mdr == "1" and flq and amg else "0"
if dst_mdr == "1" and gst_mdr == "1":
results["mdr"]["tp"].append(s)
counts["mdr"]["tp"] += 1
if dst_mdr == "0" and gst_mdr == "1":
results["mdr"]["fp"].append(s)
counts["mdr"]["fp"] += 1
if dst_mdr == "1" and gst_mdr == "0":
results["mdr"]["fn"].append(s)
counts["mdr"]["fn"] += 1
if dst_mdr == "0" and gst_mdr == "0":
results["mdr"]["tn"].append(s)
counts["mdr"]["tn"] += 1
if dst_xdr == "1" and gst_xdr == "1":
results["xdr"]["tp"].append(s)
counts["xdr"]["tp"] += 1
if dst_xdr == "0" and gst_xdr == "1":
results["xdr"]["fp"].append(s)
counts["xdr"]["fp"] += 1
if dst_xdr == "1" and gst_xdr == "0":
results["xdr"]["fn"].append(s)
counts["xdr"]["fn"] += 1
if dst_xdr == "0" and gst_xdr == "0":
results["xdr"]["tn"].append(s)
counts["xdr"]["tn"] += 1
### susceptibility
if "NA" not in [dst[s][d] for d in first_line]:
dst_sus = "1" if "1" not in [dst[s][d] for d in drugs] else "0"
gst_sus = "1" if all(
[x not in resistant_drugs for x in first_line]) else "0"
if dst_sus == "1" and gst_sus == "1":
results["sus"]["tp"].append(s)
counts["sus"]["tp"] += 1
if dst_sus == "0" and gst_sus == "1":
results["sus"]["fp"].append(s)
counts["sus"]["fp"] += 1
if dst_sus == "1" and gst_sus == "0":
results["sus"]["fn"].append(s)
counts["sus"]["fn"] += 1
if dst_sus == "0" and gst_sus == "0":
results["sus"]["tn"].append(s)
counts["sus"]["tn"] += 1
json.dump(results, open("results.json", "w"))
json.dump(counts, open("counts.json", "w"))
counts = json.load(open("counts.json"))
drugs = [x.rstrip().lower() for x in open(args.drugs).readlines()
] if args.drugs else list(counts.keys())
print("Drug\tNum\tSusceptible\tResistant\tSensitivity\tSpecificity")
for d in drugs:
if d not in counts: continue
if counts[d]["tp"] + counts[d]["fn"] == 0 or counts[d]["tn"] + counts[
d]["fp"] == 0:
continue
sensitivity = counts[d]["tp"] / (counts[d]["tp"] + counts[d]["fn"])
specificity = counts[d]["tn"] / (counts[d]["tn"] + counts[d]["fp"])
total = counts[d]["tp"] + counts[d]["fp"] + counts[d]["tn"] + counts[
d]["fn"]
suc = counts[d]["tn"] + counts[d]["fp"]
res = counts[d]["tp"] + counts[d]["fn"]
print("%s\t%s\t%s\t%s\t%s\t%s" %
(d.capitalize(), total, suc, res, sensitivity, specificity))
def profile_vcf(filename, conf):
params = conf.copy()
params["tmpvcf"] = pp.get_random_file(extension=".vcf.gz")
params["tmpcsq"] = pp.get_random_file(extension=".vcf.gz")
params["filename"] = filename
params["tmphdr"] = pp.get_random_file()
params["tmptxt"] = pp.get_random_file()
l = ""
for l in pp.cmd_out(
"bcftools view %(filename)s -h | grep \"^##FORMAT=<ID=AD\"" %
params):
pass
AD_found = False if l == "" else True
if AD_found == False:
open(params["tmphdr"], "w").write(
"##FORMAT=<ID=AD,Number=R,Type=Integer,Description=\"Allelic depths\">\n"
)
pp.run_cmd(
"bcftools query -f '%%CHROM\\t%%POS\\t%%REF\\t%%ALT\\t.[\\t0,100]\\n' %(filename)s > %(tmptxt)s"
% params)
pp.run_cmd("bgzip %(tmptxt)s" % params)
pp.run_cmd("tabix -s 1 -b 2 -p vcf %(tmptxt)s.gz" % params)
pp.run_cmd(
"bcftools view -a %(filename)s | bcftools annotate -a %(tmptxt)s.gz -c CHROM,POS,REF,ALT,-,FMT/AD -h %(tmphdr)s -Oz -o %(tmpvcf)s"
% params)
else:
pp.run_cmd("bcftools view -a %(filename)s -Oz -o %(tmpvcf)s" % params)
pp.run_cmd(
"bcftools view -T %(bed)s %(tmpvcf)s | bcftools csq -f %(ref)s -g %(gff)s -Oz -o %(tmpcsq)s -p a"
% params)
csq_bcf_obj = pp.bcf(params["tmpcsq"])
csq = csq_bcf_obj.load_csq(ann_file=conf["ann"])
results = {
"variants": [],
"missing_pos": [],
"qc": {
"pct_reads_mapped": "NA",
"num_reads_mapped": "NA"
}
}
for sample in csq:
results["variants"] = csq[sample]
all_bcf_obj = pp.bcf(params["tmpvcf"])
mutations = all_bcf_obj.get_bed_gt(conf["barcode"], conf["ref"])
if "C" in mutations["Chromosome"][325505] and mutations["Chromosome"][
325505]["C"] == 50:
mutations["Chromosome"][325505] = {"T": 25}
if "G" in mutations["Chromosome"][599868] and mutations["Chromosome"][
599868]["G"] == 50:
mutations["Chromosome"][599868] = {"A": 25}
if "C" in mutations["Chromosome"][931123] and mutations["Chromosome"][
931123]["C"] == 50:
mutations["Chromosome"][931123] = {"T": 25}
if "T" in mutations["Chromosome"][1759252] and mutations["Chromosome"][
1759252]["T"] == 50:
mutations["Chromosome"][1759252] = {"G": 25}
json.dump(mutations, open("dump.json", "w"))
barcode_mutations = pp.barcode(mutations, conf["barcode"])
results["barcode"] = barcode_mutations
results = pp.db_compare(db_file=conf["json_db"], mutations=results)
bed_regions = pp.load_bed(conf["bed"], [4], 4)
missing_regions = {gene: "NA" for gene in bed_regions}
results["missing_regions"] = missing_regions
if AD_found:
pp.run_cmd("rm %(tmpcsq)s" % params)
else:
pp.run_cmd("rm %(tmpcsq)s %(tmphdr)s %(tmptxt)s*" % params)
return results