Code accompanying The evolutionary history of 2,658 cancers

This repository contains analysis code for the aforementioned publication.

Note that some referenced files in ref and final folders are protected access and are note part of this repository.

Contents

The code folder contains a number of scripts to run MutationTime.R to estimate the timing of point mutations and copy number gains in individual samples. This step can be parallelised across individual samples.

• code/PCAWG-functions.R A collection of utility functions. This is sourced before running. This functions also sources a number of working group output files, from ref and final folder, which are not part of the repository at the moment.
• code/VCF-annotate.R An R script to annotate point mutations and estimate the timing of copy number gains. It is usually envoked via

$ bsub < VCF-annotate.sh

which submits an array job to an LSF cluster. Input to this script are (assumed in final, not provided): * consensus copy number segments * consensus subclonal reconstruction * VCF files for SNVs and indels * inferred gender

The run time ranges from 5 minutes to about 8 hours, depending on the number of variants per sample (median about 10 minutes).

• The main code file is code/PCAWG-final.R; this is run on R-3.3.3 using

> rmarkdown::render("code/PCAWG-final.R", clean=FALSE)

The output is a html report with mixed code, results and figures, which can be found at run/PCAWG-final.html. Runtime is approximately 4hr, with most time spent on loading all data. It requires about 50Gb of memory per core.

• Experimental code can be found in PCAWG-sandbox.R.
• Code used for timing simulations is in PCAWG-simulate.R.
• modules/MutationTime.R contain the MutationTimeR submodule. Don't forget to run git submodule init.

Dependencies

The main analysis workflow is run in

R-3.3.3

Additionally it requires the following R libraries

Rsamtools
VariantAnnotation
MutationTimeR (github.com/gerstung-lab/MutationTimeR
rstan
rmarkdown

Output

Main output is run/PCAWG-final.html, which contains code and results underlying Figures 1b, 2d-f, 3a,d, 5a, and 6.