Whole exome sequencing was performed as described previously (Chan et al., 2020 (link); Chang et al., 2021 (link)). Briefly, hybrid selection was done using the Human All Exon kit SureSelect Target Enrichment System (Agilent Technologies) version 6 and sequenced on the Illumina HiSeq X platform (Illumina) as paired‐end 150‐base pair reads. Read pairs were aligned to the human reference genome NCBI GRC Build 37 (hg19) using Burrows‐Wheeler Aligner (BWA MEM; Wellcome Genome Campus, Hinxton; Li & Durbin, 2009 (link)). Optical duplicates were marked with Picard followed by base score recalibration using GATK version 4.1.4 (Broad Institute) for post alignment data processing (McKenna et al., 2010 (link)). Potential germline variants were screened for by filtering for the following conditions: missense or splice site variants with mapping quality >Q20, sequencing depth >50, alternate allele depth >15, min alt fraction of 0.1. Somatic variants from the resulting normal and tumor BAM files were identified using Mutect2, and subsequently annotated and prioritized using VEP (Wellcome Genome Campus; McLaren et al., 2016 (link)). Mutational signature identification was performed using SigProfiler Bioinformatics Tools (Wellcome Genome Campus; Alexandrov et al., (link)). Copy‐number segmentations were processed with TitanCNA v1.17.1 (University of British Columbia; Ha et al., 2014 (link)).