IRCC GC Targetted NGS Protocol

This document summarizes procedures for generating 243-GEA gene panel mutation and copy number data output. Data analysis has been performed by Adam Bass informatics group at the Dana-Farber Cancer Institute (Boston, MA).

Mutation and Copy Number Alteration Analysis

DNA extracted from PDX models along with a sample of normal germline DNA from each patient were utilized for targeted next generation sequencing of a bait set of 243 genes selected based upon their alteration in prior studies of gastroesophageal cancer (the “243-GEA gene panel” was designed using SureSelect DNA platform; Agilent genomics). Additionally, hybrid capture using the human bait set followed by sequencing was performed on a NOD-SCID mouse germline DNA sample.

NGS was performed on an Illumina HiSeq2500 sequencer. 250-bp read pairs were aligned to the hg19 reference sequence using the BWA and data were sorted and duplicate-marked using Picard tools. The alignments were further refined using the Genome Analysis Toolkit (GATK) for localized realignment around indel sites . Recalibration of quality scores was also performed using the GATK.

Somatic events were identified with MuTect v1.1.4 and annotated using Variant Effect Predictor v79 (VEP). We used the SomaticIndelDetector tool that is part of the GATK for indel calling. Variants detected in PDXs that were also identified in murine DNA were filtered and removed. Only non-synonymous mutations were considered (missense, nonsense, splice donor, splice acceptor, start/stop lost, frameshift mutations, in-frame insertions/deletions).

RobustCNV was used for the identification of copy number variants. Read depth at informative capture targets in tumor samples was calibrated to estimate the copy ratio using depths observed in a panel of normal (non-cancer) diploid genomes. The copy-ratio profiles were then segmented using the circular binary segmentation (CBS) algorithm. Segments were assigned gain, loss, or normal-copy calls using a cutoff derived from the within-segment standard deviation of post-normalized mapping depths and a tuning parameter which was set based on comparisons to array-CGH calls in separate validation experiments.