def variantCalling_mutect2_partial(chromosome):
# variant caller
vcf = self.output_folder + self.sample_name + ".chr" + chromosome + ".mutect2.vcf.gz"
parameters_dict = {
"input_tumor":
self.output_folder + self.sample_name + "_r_tumor.chr" +
chromosome + "_RG.bam",
"input_normal":
self.output_folder + self.sample_name + "_r_normal.chr" +
chromosome + "_RG.bam",
"normal-sample":
self.sample_name + "_r_normal.chr" + chromosome + "_RG",
"tumor-sample":
self.sample_name + "_r_tumor.chr" + chromosome + "_RG",
"output":
vcf,
"reference":
reference_fasta,
# "--disable-read-filter":""
# "dbsnp":
# "emit-reference-confidence":"true",
# "max-alternate-alleles":"3"
}
mutect2_log = self.output_folder + self.sample_name + ".chr" + chromosome + ".Mutect2.log"
gatk_docker("gatk_mutect2", parameters_dict, mutect2_log, self.ram,
self.docker_images_dict["gatk"])
def filterMutectCalls(vcf_file):
# perform variant filtering
parameters_dict = {
"variant": vcf_file,
"output": vcf_file.replace(".vcf.gz", ".filtered.vcf.gz")
}
filter_log = self.output_folder + self.sample_name + ".filter.log"
gatk_docker("gatk_filter_mutect", parameters_dict, filter_log,
self.ram, self.docker_images_dict["gatk"])
def validateSam():
# Validate alignment file integrity
parameters_dict = {
"input":output_folder+self.sample_name+".bwa.bam",
"MODE":"SUMMARY"
}
validate_log = output_folder+self.sample_name+".validateSamFile.log"
gatk_docker(gatk_validate_sam, parameters_dict,
HaplotypeCaller_log, self.ram,self.docker_images_dict["gatk"])
def buildRecalibrator():
# Base recalibrator (GATK)
parameters_dict = {
"input":sorted_bam,
"output":BaseRecalibrator_metrics,
"reference":reference_fasta,
"known-sites":dbsnp_vcf,
"use-original-qualities":"true"
}
BaseRecalibrator_log = output_folder+self.sample_name+".BaseRecalibrator.log"
gatk_docker("gatk_build_recalibrator", parameters_dict,
BaseRecalibrator_log, self.ram,self.docker_images_dict["gatk"])
def variantCalling_HaplotypeCaller(input_bam):
# Haplotype caller
parameters_dict = {
"input": input_bam,
"output": input_bam.replace(".bam", ".vcf.gz"),
"reference": reference_fasta
# "emit-reference-confidence":"true",
# "max-alternate-alleles":"3"
}
HaplotypeCaller_log = input_bam + ".HaplotypeCaller.log"
gatk_docker("gatk_haplotype_caller", parameters_dict,
HaplotypeCaller_log, self.ram,
self.docker_images_dict["gatk"])
def addReadGroups():
# Add read groups (GATK)
parameters_dict = {
"input":unsorted_bam,
"output":ReadGroups_bam,
"RGLB":self.lib_ID,
"RGPL":self.pl_ID,
"RGPU":self.pu_ID,
"RGSM":self.sample_name
}
ReadGroups_log = self.output_folder+self.sample_name+".ReadGroups.log"
gatk_docker("gatk_add_read_groups", parameters_dict,
ReadGroups_log, self.ram,self.docker_images_dict["gatk"], )
def addReadGroups(t_n):
# Add read groups (GATK)
parameters_dict = {
"input": markDuplicates_bam,
"output": ReadGroups_bam,
"RGLB": self.lib_ID,
"RGPL": self.pl_ID,
"RGPU": self.pu_ID,
"RGSM": self.sample_name + "_r_" + t_n
}
ReadGroups_log = self.output_folder + t_n_sample_name + ".ReadGroups." + t_n + ".log"
gatk_docker("gatk_add_read_groups", parameters_dict,
ReadGroups_log, self.ram,
self.docker_images_dict["gatk"])
def addReadGroups_partial_bam(chromosome, input_bam):
# Add read groups (GATK)
parameters_dict = {
"input": input_bam,
"output": input_bam.replace(".bam", "_RG.bam"),
"RGLB": self.lib_ID,
"RGPL": self.pl_ID,
"RGPU": self.pu_ID,
"RGSM": input_bam.replace(".bam", "_RG").split("/")[-1]
}
ReadGroups_log = self.output_folder + t_n_sample_name + ".ReadGroups.log"
gatk_docker("gatk_add_read_groups", parameters_dict,
ReadGroups_log, self.ram,
self.docker_images_dict["gatk"])
def applyRecalibrator():
# Base recalibrator - applying model (GATK)
parameters_dict = {
"input":sorted_bam,
"output":BaseRecalibrator_bam,
"bqsr":BaseRecalibrator_metrics,
"use-original-qualities":"true",
"static-quantized-quals":"10",
"static-quantized-quals":"20",
"static-quantized-quals":"30"
}
ApplyBQSR_log = output_folder+self.sample_name+".ApplyBQSR.log"
gatk_docker("gatk_apply_recalibrator", parameters_dict,
ApplyBQSR_log, self.ram,self.docker_images_dict["gatk"])
def markDuplicates(input_bam):
# Mark duplicates (GATK)
parameters_dict = {
"input": input_bam,
"output": input_bam.replace("bam", "MarkDuplicates.bam"),
"mark_dupl_metrics": markDuplicates_metrics,
"optical_duplicate_pixel_dist": "2500",
"assume_sort_order": "queryname",
"clear_DT": "false",
"add_pg_tag_to_reads": "false"
}
markDuplicates_log = self.output_folder + self.sample_name + ".MarkDuplicates.log"
gatk_docker("gatk_mark_duplicates", parameters_dict,
markDuplicates_log, self.ram,
self.docker_images_dict["gatk"])
def addReadGroups():
# Add read groups (GATK)
print("")
print("STARTING ADDING READGROUPS")
print("")
parameters_dict = {
"input": unsorted_bam,
"output": ReadGroups_bam,
"RGLB": self.lib_ID,
"RGPL": self.pl_ID,
"RGPU": self.pu_ID,
"RGSM": self.sample_name
}
ReadGroups_log = self.output_folder + self.sample_name + ".ReadGroups.log"
gatk_docker("gatk_add_read_groups", parameters_dict,
ReadGroups_log, self.ram,
self.docker_images_dict["broadinstitute/gatk"])
