Somatic mutation data (hg38 reference genome)

Annotated .maf somatic mutation data files for 30 tumour samples from 22 patients Method for somatic mutation calling Somatic single nucleotide variants (SNVs), insertions and deletions (InDels) were called using Mutect2 (v.4.1.8) and Strelka (v. 2.9.10) respectively from matched normal and tumour pairs. Strelka was run with the --exome option (for WXS data only) and –callRegions option to restrict mutation calling to chr1-22,X,Y,M. In order to filter for false positive somatic mutation calls such as common variants and mapping artifacts, Mutect2 was run with gnomAD germline population reference and a panel of normal (PON) samples, generated using the CreateSomaticPanelOfNormals function part of the GATK4 (v.4.1.8) best practise pipeline. FFPE samples are known to contain mutational biases in the C>T/G>A transition. OxoG filter was applied through the read orientation bias model with Mutect2 to remove mutations with FFPE strand bias. GATK4 GetPileupSummaries and CalculateContamination was used with a set number of known germline common variants reported in ExAC at a population minor allele frequency > 0.05 to calculate cross sample contamination. FilterMutectCalls was run using default parameters. Filtered Mutect2 and Strelka somatic variant calls were combined into one vcf using GATK3 (v.3.8.1) CombineVariants. Bcftools (v.1.12) (http://samtools.github.io/bcftools/bcftools.html) norm function was used to left align and normalise InDels. Variants passing quality control were annotated using MSK vcf2maf (https://github.com/mskcc/vcf2maf) and variant effect predictor (VEP v.96) using GRCh38, which outputs both a .vcf and .maf file format. Annotated maf files were used for downstream somatic mutation analysis.