Benchmarking somatic variant calling using multi-regional sequencing data

This repository contains workflow and analysis scripts pertaining to a simulation study that aims to assess the performance of several popular somatic variant calling methods.

The study was carried out using two different simulation approaches:

De-novo - Simulated Reads, Simulated Variants (full simulation)

1. Reads were simulated from an artificial reference genome.
2. Germline and somatic variants were simulated and spiked into the reads.

Spike-in - Real Reads, Simulated Variants (partial simulation)

1. An empirical sequencing data set was split into multiple samples.
2. Reads were sorted by germline allele (maternal/paternal allele) using germline variants.
3. Somatic variants were simulated and spiked into the reads.

Structure of this repository

• R code to generate the plots for the publictation are in the scripts de-novo.R and spike-in.R
• Scripts that show how variant callers where run are in the respective directories under workflow.
• Technical details about the simulations can be gleaned from the content of the sim directories.