Module 8: Somatic Mutation Discovery: GATK Mutect2 & FilterMutectCalls

1. Running Mutect2 on Targeted Panel Regions

Having produced coordinate-sorted, duplicate-marked BAM files, we now execute GATK Mutect2 restricted to the targeted capture territory (target_regions.bed) to detect somatic SNVs and indels:

Galaxy Parameter Configuration for GATK4 Mutect2:
- Reference Genome: Human (GRCh38/hg38)
- Tumor BAM: Case1_Lung_Tumor_Marked.bam (Sample Name: Case1_Tumor)
- Normal BAM: Case1_Lung_Normal_Marked.bam (Sample Name: Case1_Normal)
- Interval list (-L): Lung_10Gene_Panel_Targets.bed
- Germline Resource: gnomAD Somatic AF-only VCF (v3.1)
- Panel of Normals (PoN): Curated_CLIA_PoN_GRCh38.vcf.gz
- Output: Raw Somatic VCF + F1R2 orientation metrics

2. Estimating Cross-Sample Contamination: GetPileupSummaries & CalculateContamination

In high-throughput multi-plexed clinical sequencing, index hopping and liquid handling cross-contamination can introduce traces of DNA from other patient samples. GATK calculates contamination by examining population-polymorphic biallelic SNPs:

1. GetPileupSummaries: Quantifies reference and alternate depths across thousands of known common SNPs (gnomAD AF 5% – 20%).
2. CalculateContamination: Computes the maximum likelihood estimate of cross-sample contamination ($f_{\text{contam}}$). For clinical cancer specimens, contamination must remain strictly $< 1.0\% - 2.0\%$.

3. Rigorous Filtering with FilterMutectCalls

The raw VCF produced by Mutect2 contains thousands of low-confidence candidate calls. FilterMutectCalls evaluates multi-dimensional filters:

Filter the final dataset to retain only variants exhibiting FILTER == PASS with total depth $\ge 100 imes$ and VAF $\ge 5\%$.