Cholangiocarcinoma mutational sources and selection

We aggregated somatic mutation data from 14 sources and analyzed mutational signatures and somatic selection. Starting from MAF files, all data and scripts necessary to reproduce analyses and figures are included here.

Key contents

Sample information

• combined_sample_key.txt: Contains all sample-level information used in the analyses.
• study_sample_keys/: Sample information by study. Some of the study-specific files have additional information that wasn't used in analyses.
• maf_file_summary.txt: Information about all final MAF files used in analyses.
• data_source_summary.txt: Information about data sources, how sequencing was performed in each, and links to related publications.

Data and pre-processing

• data_sources/: Contains one subdirectory per data source. For each source, there is original MAF data (as acquired from sources, or generated with our GATK/Mutect2 pipeline; see Methods), processing scripts, and a processed MAF file.
• data_prep/: Final data pre-processing script that creates harmonized MAF files; some ancillary information.
• final_mafs/: The harmonized MAF files used in all analyses.
• targeted_regions/: BED files specifying the targeted regions of all exome capture arrays and gene sequencing panels. See methods for details.
• reference/: Additional external data needed for analyses, including MSigDB-derived pathway definitions and UCSC chain files for genome coordinate conversion.

Analyses

• output/: Files generated by analysis scripts, including numerical results. Delete or rename to reproduce results.
• analyses/: Scripts used for the analyses. Scripts use data saved to the output directory. To replicate analyses, remove previous outputs and run main_analysis.R. From there, most analyses can be run in any order. Subdirectories are numbered to indicate when scripts must be run in a certain order.
• figures/: Scripts for making figures from analysis output, as well as some intermediary files.
• final_figures/: Final copies of main figures.

Software prerequisites

All analyses should be runnable using a recent version of R. You will need the cancereffectsizeR package; for precise replication, use version 2.10.2.