Workflow 3: sample level variant calling

This workflow will call variants for the samples including:

• sample level recalibration
• sample level realignment

N.B. no sample level MarkDuplicates? is needed as lanes = libraries.

Workflow inputs:

• lane.chr.recal.sorted.bam - for all sample lanes: dedupped, recalibrated, realigned, sorted and indexed bams (3)
• sample.chip.vcf - genotypes called from genotype chip

Reference:

• genome.chr.fasta - reference genome split on chromosome
• genome.chr.realign.intervals - targets for realignment per chromosome
• genome.chr.dbsnpXYZ.rod - known snp variants, here from dpbsnp
• genome.chr.indelsXYZ.vcf - known indels from, here from 1KG

Workflow outputs:

• sample.chr.bam - merged bam files per sample
• sample.chr.realign.interval - realignment target intervals
• sample.chr.realigned.bam - realigned
• sample.chr.matesfixed.bam - fixed pairs in realignment
• sample.chr.indels.vcf - raw indels called
• sample.chr.indels.bed - raw indels annotations
• sample.chr.indels.txt - output from the indel calling
• sample.chr.indels.filtered.bed - indels filtered
• sample.chr.snps.vcf - raw snps called
• sample.chr.snps.filtered.vcf - snps filtered

merge-lanes

Merge lanes into one sample bam

tool:	sam merge
inputs:	lane.chr.recal.sorted.bam
outputs:	sample.chr.bam
docs:	​http://samtools.sourceforge.net/samtools.shtml

RealignerTargetCreator

Create realignment targets based on the data (so not only knowns)

tool:	GenomeAnalysisTK.jar -T RealignerTargetCreator?
inputs:	sample.chr.bam genome.chr.fa dbsnpXYz.chr.rod indelsXYZ.vcf
outputs:	sample.chr.realign.intervals
doc:	​http://www.broadinstitute.org/gsa/wiki/index.php/Local_realignment_around_indels#Creating_Intervals

IndelRealigner

Realign based on realignment targets in previous step

tool:	GenomeAnalysisTK.jar -T IndelRealigner?
inputs:	sample.chr.bam genome.chr.realign.intervals genome.chr.dbsnpXYZ.rod genome.chr.indelsXYZ.vcf
outputs:	sample.chr.realigned.bam
doc:	​http://www.broadinstitute.org/gsa/wiki/index.php/Local_realignment_around_indels#Realigning

FixMateInformation

See description in workflow2, now applied to sample

inputs:	sample.chr.realigned.bam
ouputs:	sample.chr.matesfixed.bam

IndelGenotyperV2

Call indels

tool:	GenomeAnalysisTK.jar -T IndelGenotyperV2
inputs:	sample.chr.matesfixed.bam genome.chr.fa
outputs:	sample.chr.indels.vcf sample.chr.indels.bed sample.chr.indels.txt
doc:	​http://www.broadinstitute.org/gsa/wiki/index.php/Indel_Genotyper_V2.0

​http://www.broadinstitute.org/gsa/wiki/index.php/Firehose_Parameters#SampleIndelGenotyper

filterSingleSampleCalls

Filter indels

tool:	filterSingleSampleCalls.pl
inputs:	sample.chr.indels.bed
outputs:	sample.chr.indels.filtered.bed
doc:	​http://www.broadinstitute.org/gsa/wiki/index.php/Firehose_Parameters#SampleIndelGenotyper

UnifiedGenotyper

Call SNPs

tool:	GenomeAnalysisTK.jar -T UnifiedGenotyper?
inputs:	sample.chr.matesfixed genome.chr.fa dbsnpXYz.chr.rod
outputs:	sample.chr.snps.vcf
doc:	​http://www.broadinstitute.org/gsa/wiki/index.php/Firehose_Parameters#SetUnifiedGenotypertoEval

​http://www.broadinstitute.org/gsa/wiki/index.php/Unified_genotyper

makeIndelMask

Make indel mask

tool:	makeIndelMask.py
inputs:	sample.chr.indels.bed
outputs:	sample.chr.indels.mask.bed
doc:	​http://www.broadinstitute.org/gsa/wiki/index.php/Indel_Genotyper_V2.0#Creating_a_indel_mask_file

VariantFiltration

Filter variants to get the best calls possible

tool:	GenomeAnalysisTK.jar -T VariantFiltration?
inputs:	sample.chr.snps.vcf genome.chr.fa dbsnpXYz.chr.rod
outputs:	sample.chr.snps.filtered.vcf
doc:	​http://www.broadinstitute.org/gsa/wiki/index.php/Best_Practice_Variant_Detection_with_the_GATK_v2#Integrating_analyses:_getting_the_best_call_set_possible

MergeVcfs

ChipVcf

Produce vcf for the chips

VariantEval

Create summary information on the variations called for evaluation. Run per sample.snps.filtered.vcf against chip.

tool:	GenomeAnalysisTK.jar -T VariantEval?
inputs:	sample.snps.vcf sample.chip.vcf genome.chr.fa dbsnpXYz.chr.rod
outputs:	sample.snps.eval
doc:	​http://www.broadinstitute.org/gsa/wiki/index.php/VariantEval

Discussion:

Do we call SNPs based on the filtered indels or the raw indels? Should we realign AGAIN after merge of lanes? BAQ? MINDEL/PINDEL?