Somatic Variant Calling from Exome Sequencing

Genomic DNA was extracted from whole bone marrow aspirate samples using Autopure Extractor (QIAGEN). Exome capture hybrid was performed with using Agilent SureSelect All Exon V4. Illumina HiSeq 2000 sequencer was used for sequencing with 75 base pair paired end read. Sequencing data was aligned to the hg19 human genome reference using Burrows-Wheeler Aligner (BWA) [33 (link)] followed by mark duplication, indel realignment, and base recalibration using Genome Analysis Toolkit (https://www.broadinstitute.org/gatk/guide/best-practices?bpm=DNAseq). The resulting BAM [34 (link)] files were preprocessed and base substitutions and small insertions/deletions were called using Mutect [35 (link)] and Pindel [36 (link)], respectively. To overcome the lack of matched normal sample, we generated virtual common normal sequence using in-house pooled normal sequence. This method has been shown to function as almost equivalent as matched normal to call somatic variants [37 (link)]. Since we intended to use mutation data for potential clinical application, we wanted to be conservative on variant annotation. Therefore, we modified approach used by Pappaemanuil et al. (5) to call high-confidence driver mutations (Supplementary Methods).