def filterVCF(vcf, SAMPLE_NUM):
output_path = os.path.join(OUT_DIR, "filtered.vcf.gz_{0}".format(SAMPLE_NUM))
cmd = [GATK, "VariantFiltration", "-R", REF, "-V", vcf, "-O", output_path]
stdout = runCMD(cmd)
return output_path
def splitNCigarReads(bam, SAMPLE_NUM):
output_path = os.path.join(OUT_DIR, "splitted.bam_{0}".format(SAMPLE_NUM))
cmd = [GATK, "SplitNCigarReads", "-R", REF, "-I", bam, "-O", output_path]
stdout = runCMD(cmd)
return output_path
def haploCaller(bam, SAMPLE_NUM):
output_path = os.path.join(OUT_DIR, "output.g.vcf.gz_{0}".format(SAMPLE_NUM))
cmd = [GATK, "HaplotypeCaller", "-R", REF, "-I", bam, "-O", output_path, "-ERC", "GVCF", "-L", "chr1", "-L", "chr2"]
stdout = runCMD(cmd)
return output_path
def convertSAMtoFASTQ(usam, SAMPLE_NUM):
output_path = os.path.join(OUT_DIR, "sample.fastq_{0}".format(SAMPLE_NUM))
input_arg = "I={0}".format(usam)
output_arg = "FASTQ={0}".format(output_path)
cmd = ["java", "-jar", PICARD, "SamToFastq", input_arg, output_arg]
stdout = runCMD(cmd)
return output_path
def bwa_map(fastq, SAMPLE_NUM):
output_path = os.path.join(OUT_DIR, "mapped.sam_{0}".format(SAMPLE_NUM))
cmd = [BWA, "mem", REF, fastq]
f = open(output_path, "w")
stdout = runCMD(cmd, f)
f.close()
return output_path
def revertSAM(ubam, SAMPLE_NUM):
output_path = os.path.join(OUT_DIR, "reverted.bam_{0}".format(SAMPLE_NUM))
input_arg = "I={0}".format(ubam)
output_arg = "O={0}".format(output_path)
cmd = ["java", "-jar", PICARD, "RevertSam", input_arg, output_arg]
stdout = runCMD(cmd)
return output_path
def convertSAMtoBAM(sam, SAMPLE_NUM):
output_path = os.path.join(OUT_DIR, "mapped.bam_{0}".format(SAMPLE_NUM))
cmd = [SAMTOOLS, "view", "-bhS", sam]
f = open(output_path, "w")
stdout = runCMD(cmd, f)
f.close()
return output_path
def analyzeCovariates(recal_data, SAMPLE_NUM):
output_path = os.path.join(OUT_DIR,
"AnalyzeCovariates.pdf_{0}".format(SAMPLE_NUM))
cmd = [
GATK, "AnalyzeCovariates", "-bqsr", recal_data, "-plots", output_path
]
stdout = runCMD(cmd)
return output_path
def annotateVCF(bam, vcf, SAMPLE_NUM):
output_path = os.path.join(OUT_DIR, "annotated.vcf_{0}".format(SAMPLE_NUM))
cmd = [
GATK, "VariantAnnotator", "-I", bam, "-R", REF, "-V", vcf, "-O",
output_path, "-A", "Coverage", "--dbsnp", DBSNP
]
stdout = runCMD(cmd)
return output_path
def sort(bam, SAMPLE_NUM):
output_path = os.path.join(OUT_DIR, "sorted.bam_{0}".format(SAMPLE_NUM))
output_arg = "O={0}".format(output_path)
input_arg = "I={0}".format(bam)
sort_order = "SORT_ORDER=queryname"
cmd = [
"java", "-jar", PICARD, "SortSam", input_arg, output_arg, sort_order
]
stdout = runCMD(cmd)
return output_path
def applyBQSR(bam, recal_data, SAMPLE_NUM):
output_path = os.path.join(OUT_DIR,
"recalibrated.bam_{0}".format(SAMPLE_NUM))
cmd = [
GATK, "ApplyBQSR", "-R", REF, "-I", bam,
"--add-output-sam-program-record", "--use-original-qualities", "-O",
output_path, "--bqsr-recal-file", recal_data
]
stdout = runCMD(cmd)
return output_path
def recalibrateBase(bam, SAMPLE_NUM):
output_path = os.path.join(OUT_DIR,
"recal_data.table_{0}".format(SAMPLE_NUM))
known_sites1 = DBSNP
cmd = [
GATK, "BaseRecalibrator", "-I", bam, "-R", REF, "--known-sites",
known_sites1, "-O", output_path, "-L", "chr1", "-L", "chr2"
]
stdout = runCMD(cmd)
return output_path
def mergeBamAlignment(mapped_bam, unmapped_sam, SAMPLE_NUM):
output_path = os.path.join(OUT_DIR, "merged.bam_{0}".format(SAMPLE_NUM))
output_arg = "O={0}".format(output_path)
input_arg1 = "ALIGNED={0}".format(mapped_bam)
input_arg2 = "UNMAPPED={0}".format(unmapped_sam)
ref = "R={0}".format(REF)
cmd = [
"java", "-jar", PICARD, "MergeBamAlignment", input_arg1, input_arg2,
ref, output_arg
]
stdout = runCMD(cmd)
return output_path
def markDuplicates(mapped_bam, SAMPLE_NUM):
output_path1 = os.path.join(OUT_DIR,
"marked_duplicates.bam_{0}".format(SAMPLE_NUM))
output_arg = "O={0}".format(output_path1)
output_path2 = os.path.join(OUT_DIR, "marked_dup_metrics")
output_arg2 = "M={0}".format(output_path2)
input_arg = "I={0}".format(mapped_bam)
cmd = [
"java", "-jar", PICARD, "MarkDuplicates", input_arg, output_arg,
output_arg2
]
stdout = runCMD(cmd)
return output_path1, output_path2
def run(self, buildType):
tCompCmd = self._formatCompileCMD()
try:
tCompileLog = self._getLogFileName()
with open(tCompileLog, "w+") as tFile:
compileOut = runCMD(tCompCmd, workDir=self._getRunDir(), pipe=tFile,
isShell=platform.system() is not "Windows")
except:
log.error("[Error] Can't compile project: {0}".format(sys.exc_info()[1]))
return False
else:
log.info("[Info] Compile log saved to: {0}".format(tCompileLog))
if compileOut["ret_code"] != 0:
log.error("[Error] {0} Compiler log {1}".format("=" * 20, "=" * 20))
self._printNiceFailLog(compileOut["out"])
log.error("[Error] {0}".format("-" * 51))
return False
else:
return True
def addOrReplaceReadGroups(bam, SAMPLE_NUM):
output_path = os.path.join(OUT_DIR, "output_rg.bam_{0}".format(SAMPLE_NUM))
output_arg = "O={0}".format(output_path)
input_arg = "I={0}".format(bam)
rgid = "RGID=4"
rglb = "RGLB=lib1"
rgpl = "RGPL=ILLUMINA"
rgpu = "RGPU=unitl"
rgsm = "RGSM=20"
cmd = [
"java", "-jar", PICARD, "AddOrReplaceReadGroups", input_arg,
output_arg, rgid, rglb, rgpl, rgpu, rgsm
]
stdout = runCMD(cmd)
indexBAM(output_rg)
return output_path
def run(self, buildType):
assert self._rootPath is not None # check that root path specified
self._setUpBuildInfo(buildType)
cmakeCMD = "cmake -H." + self._getBuildDirCMD() + self._getGenCMD() + self._getDefsCMD()
with open(self._getLogFile(), "w+") as tFile:
setUpEnv(self._getEnvVars())
try:
tProcessOut = runCMD(cmakeCMD, workDir=self._getRunDir(), pipe=tFile,
isShell=platform.system() is not "Windows")
except:
log.error("[Error] Can't run cmake: {0} ".format(sys.exc_info()[1]))
return None
else:
log.info("[Info] Cmake log saved to: {0}".format(self._getLogFile()))
if tProcessOut["ret_code"] != 0:
log.error("[Error] {0} Cmake log {1}".format("=" * 18, "=" * 19))
self._printNiceFailLog(tProcessOut["out"])
log.error("[Error] {0}".format("-" * 51))
return None
else:
return self._buildDict
def _checkToolVersion(self, binName, matchInfo):
tCmdArgs = [binName, matchInfo["args"]]
# Uncoment for easy debug
#return True
tProcOut = runCMD(tCmdArgs, isShell=platform.system().lower() == "windows")
#return True
if tProcOut["ret_code"] != 0:
log.error("[Error] Can't get version of: {0}".format(binName))
return False
tLines = tProcOut["out"].split("\n")
if len(tLines) < matchInfo["line"]:
log.error("[Error] Invalid number of lines in output of: {0}".format(binName))
log.error("[Error] Expected count of lines: {0}".format(matchInfo["line"]))
return False
tWords = tLines[matchInfo["line"]].split(" ")
if len(tWords) < matchInfo["word"]:
log.error("[Error] Invalid count of word ({0} but need {1}): {2}"
.format(len(tWords), matchInfo["word"], binName))
return False
tPhrase = tWords[matchInfo["word"]]
matchInfo["actual_ver"] = tPhrase
return self._isValidVersion(tPhrase, matchInfo["match"])
def indexBAM(bam):
cmd = [SAMTOOLS, "index", bam]
stdout = runCMD(cmd)