def diffvcf(self):
# Description for duos usage
try:
patient1 = argv[2]
patient2 = argv[3]
# Checks file existence and type for patients
pt1Check = checkFile(patient1, ".vcf")
pt2Check = checkFile(patient2, ".vcf")
logger.info(
"Evaluating differences between %s and %s"
% (str(patient1), str(patient2))
)
pvcfs = vcfmgr.ParsedVCF.mp_parser(patient1, patient2)
result = pvcfs[0].duos(pvcfs[1])
resultname = result.name
outpath = (
cfg.resultsPath
+ "Diffs/"
+ result.name.replace(":", "_")
+ "/"
+ result.name.replace(":", "_")
)
result.name = resultname
outpath = outpath + ".xlsx"
logger.info("Writing Result File")
result.vcf_to_excel(outpath)
logger.info("Diff Analisis Complete")
logger.info("File available at:%s" % outpath)
except Exception as e:
logger.info("Diff Analisis Failed")
logger.error(str(e))
return 0
def single(patient, panel):
try:
checkFile(patient, ".vcf")
checkFile(panel, ".xlsx")
logger.info("Running %s on patient %s" % (str(panel), str(patient)))
result = vcfmgr.ParsedVCF.from_vcf(patient).panel(panel)
outpath = (cfg.patientPath + result.name + "/Panels/" + result.name +
"_" + panel + ".xlsx")
os.makedirs(os.path.dirname(outpath), exist_ok=True)
result.vcf_to_excel(outpath)
logger.info("Single Analisis Complete")
logger.info("File available at:%s" % outpath)
return outpath
except Exception as err:
logger.error("Single analysis Failed")
logger.debug(f"Error was: {err}", exc_info=True)
raise Exception
def trios(patient1,
patient2,
patient3,
VennPlace=None,
Filter=None,
Panel=None):
try:
checkFile(patient1, ".vcf")
checkFile(patient2, ".vcf")
checkFile(patient3, ".vcf")
logger.info("Running Trios Study on %s, %s and %s" %
(str(patient1), str(patient2), str(patient3)))
pvcfs = vcfmgr.ParsedVCF.mp_parser(patient1, patient2, patient3)
result = pvcfs[0].duos(pvcfs[1]).duos(pvcfs[2], VENNPLACE=VennPlace)
resultname = result.name
outpath = (cfg.resultsPath + "Trios/" + result.name.replace(":", "_") +
"/" + result.name.replace(":", "_"))
result.name = resultname
if VennPlace is not None:
outpath = outpath + "_Venn" + VennPlace.replace(":", "_")
# check if there is a Panel Requested
if Panel:
logger.info("Running panel {}".format(Panel))
panel = Panel
checkFile(panel, ".xlsx")
result = result.panel(panel)
result.name = resultname
outpath = outpath + "_Panel" + Panel
# check if there is a Filter Requested
if Filter[0] is not None:
for x in Filter:
if (len(x.split())) != 2:
logger.error("--Filter accepts only two arguments. \
Usage: --Filter COLUMN_NAME TEXT_TO_FILTER")
exit(1)
else:
x = x.split()
if x[1] == "Empty":
result = result[result[x[0]] != ""]
else:
result = result[~result[x[0]].str.contains(x[1])]
result.name = resultname
outpath = outpath + "_Filter" + str(x[0]) + str(x[1])
outpath = outpath + ".xlsx"
result.vcf_to_excel(outpath)
logger.info("Trios Analisis Complete")
logger.info("File available at:%s" % outpath)
except Exception as err:
logger.error("Trios Analisis Failed")
logger.debug(f"Error was: {err}", exc_info=True)
raise Exception
def single(self):
# Description for panel usage
parser = argparse.ArgumentParser(description="Run study on a single patient")
parser.add_argument(
"-Panel", required=True, help="File name of Panel inside Panels folder"
)
parser.add_argument(
"-Patient",
required=True,
help="Patient File Path - It needs to match exactly to the one found inside Patients folder",
)
# ignore first argument
try:
args = parser.parse_args(argv[2:])
panel = cfg.panelsPath + args.Panel + ".xlsx"
patient = cfg.patientPath + args.Patient
ptCheck = checkFile(patient, ".vcf")
pnCheck = checkFile(panel, ".xlsx")
logger.info(
"Running %s on patient %s" % (str(args.Panel), str(args.Patient))
)
result = vcfmgr.ParsedVCF.from_vcf(patient).panel(panel)
outpath = (
cfg.patientPath
+ result.name
+ "/Panels/"
+ result.name
+ "_"
+ args.Panel
+ ".xlsx"
)
os.makedirs(os.path.dirname(outpath), exist_ok=True)
result.vcf_to_excel(outpath)
logger.info("Single Analisis Complete")
logger.info("File available at:%s" % outpath)
except:
logger.info("Single Analisis Failed")
return 0
def trios(self):
parser = argparse.ArgumentParser(description="Run Trios Study on two patients")
parser.add_argument(
"-Patient1",
required=True,
help="Patient 1 File Path - It needs to match exactly to the one found inside Patients folder",
)
parser.add_argument(
"-Patient2",
required=True,
help="Patient 2 File Path - It needs to match exactly to the one found inside Patients folder",
)
parser.add_argument(
"-Patient3",
required=True,
help="Patient 3 File Path - It needs to match exactly to the one found inside Patients folder",
)
parser.add_argument(
"--Panel", nargs="?", const=None, help="Panel to run on Trios study"
)
parser.add_argument(
"--Filter",
nargs="?",
const=None,
help="Filter to apply. This function will filter out every row that includes the given text"
" in the given column. For filtering Empty data, TEXT keyword is 'Empty'",
metavar=("COLUMN TEXT"),
action="append",
)
parser.add_argument(
"--VennPlace",
default=None,
const=None,
nargs="?",
choices=["A", "B", "C", "A:B", "A:C", "B:C", "A:B:C", "ALL"],
help="Place in a Venn Diagram to obtain variants from",
)
try:
# ignore first argument
args = parser.parse_args(argv[2:])
patient1 = cfg.patientPath + args.Patient1
patient2 = cfg.patientPath + args.Patient2
patient3 = cfg.patientPath + args.Patient3
# Checks file existence and type for patients
pt1Check = checkFile(patient1, ".vcf")
pt2Check = checkFile(patient2, ".vcf")
pt3Check = checkFile(patient3, ".vcf")
logger.info(
"Running Trios Study on %s, %s and %s"
% (str(args.Patient1), str(args.Patient2), str(args.Patient3))
)
pvcfs = vcfmgr.ParsedVCF.mp_parser(patient1, patient2, patient3)
result = pvcfs[0].duos(pvcfs[1]).duos(pvcfs[2], VENNPLACE=args.VennPlace)
resultname = result.name
outpath = (
cfg.resultsPath
+ "Trios/"
+ result.name.replace(":", "_")
+ "/"
+ result.name.replace(":", "_")
)
result.name = resultname
if args.VennPlace is not None:
outpath = outpath + "_Venn" + args.VennPlace.replace(":", "_")
# check if there is a Panel Requested
if args.Panel:
logger.info("Running panel {}".format(args.Panel))
panel = cfg.panelsPath + args.Panel + ".xlsx"
checkFile(panel, ".xlsx")
result = result.panel(panel)
result.name = resultname
outpath = outpath + "_Panel" + args.Panel
# check if there is a Filter Requested
if args.Filter[0] is not None:
for x in args.Filter:
if (len(x.split())) != 2:
logger.error(
"--Filter accepts only two arguments. Usage: --Filter COLUMN_NAME TEXT_TO_FILTER"
)
exit(1)
else:
x = x.split()
if x[1] == "Empty":
result = result[result[x[0]] != ""]
else:
result = result[~result[x[0]].str.contains(x[1])]
result.name = resultname
outpath = outpath + "_Filter" + str(x[0]) + str(x[1])
outpath = outpath + ".xlsx"
result.vcf_to_excel(outpath)
logger.info("Trios Analisis Complete")
logger.info("File available at:%s" % outpath)
except:
logger.info("Trios Analisis Failed")
return 0
def pipeline(self):
# Description for pipeline usage
parser = argparse.ArgumentParser(description="Run a Pipeline from FASTQ to VCF")
parser.add_argument(
"-Pipeline",
required=True,
help="File name of the Pipeline inside Pipelines folder",
)
parser.add_argument(
"-FQ",
required=True,
help="Patient FastQ1 File Path - It needs to match exactly "
"the filename found inside Patients folder. Only this one is needed for Single End."
"Two FastQs will be needed for Paired End (usage: -FQ Fastq1 -FQ Fastq2",
action="append",
)
parser.add_argument(
"-keeptmp",
action="store_true",
default=False,
help="Keep Temp files, otherwise just creates annotated vcf file.",
)
parser.add_argument(
"-startStep",
default=0,
type=int,
help="Defines step to start running pipeline.",
)
parser.add_argument(
"-endStep",
default=0,
type=int,
help="Defines step to start running pipeline.",
)
# ignore first argument
args = parser.parse_args(argv[2:])
pipe = cfg.pipelinesPath + args.Pipeline
checkFile(pipe, args.Pipeline.split(".")[-1])
newpipe = pipeline.Pipeline.from_json(pipe)
if len(args.FQ) > 2:
logger.error(
"Only Two FASTQ files allowed. The Input for FastQ Files was: ",
str(args.FQ),
)
return exit(1)
elif len(args.FQ) == 2:
fq1 = cfg.patientPath + args.FQ[0]
fq2 = cfg.patientPath + args.FQ[1]
checkFile(fq1, "." + fq1.split(".")[-1])
checkFile(fq2, "." + fq2.split(".")[-1])
if args.keeptmp:
newpipe.runpipeline(
fq1,
fq2,
keeptmp=True,
startStep=args.startStep,
endStep=args.endStep,
)
else:
newpipe.runpipeline(
fq1, fq2, startStep=args.startStep, endStep=args.endStep
)
return 0
else:
fq1 = cfg.patientPath + args.FQ[0]
fq2 = ""
checkFile(fq1, "." + fq1.split(".")[-1])
if args.keeptmp:
newpipe.runpipeline(
fq1, keeptmp=True, startStep=args.startStep, endStep=args.endStep
)
else:
newpipe.runpipeline(fq1, startStep=args.startStep, endStep=args.endStep)
return 0