A repository associated with Attribution of Cancer Origins to Endogenous, Exogenous, and Preventable Mutational Processes, Molecular and Biological Evolution (Cannataro, Mandell, and Townsend 2023).

New: The cancereffectsizeR package now has a built-in function to attribute cancer effects to mutational processes using our method. Naturally, for your own analyses, you should use this actively maintained package rather than extracting code from this repository. :)

The latest iteration of our manuscript uses a bootstrap approach to resample variants within tumors, and subsequently generate confidence intervals around detected signatures and downstream analyses. The manuscript figures and analyses are generated within manuscript_analysis_bootstrap/MAIN_script_attribution_effect_size.R. This script sources various R scripts in the R subdirectory to generate perform calculations and generate figures. It also calls input data from bootstrap_analysis/, which were generated on the Yale HPC clusters. The scripts to generate these data, gather them, and retrieve them are stored within scripts_for_cluster/with_bootstrap

Before bootstrap approach:

The RMarkdown script Population_scaled_analysis_manuscript_calculations.Rmd contains the code to calculate our main findings and generate all of the figures in our manuscript.

Analyses that generate input to Population_scaled_analysis_manuscript_calculations.Rmd were run on the high-performance computing cluster, for specific files and parameters used, see scripts_for_cluster/selection_and_pop_level_cluster_submitter_batch.sh. This shell script ultimately calls scripts_for_cluster/selection_and_pop_cluster.R, which performs the effect size, mutational source effect size, and Jensen-Shannon Divergence calculations.