Dirichlet Allocation of MUTAtions in cancer

Damage and Misrepair Signatures: Compact Representations of Pan-cancer Mutational Processes

---

Features

• Separately model damage and misrepair processes
• Estimate activities of DAMUTA signatures
• Fit new Damage- and Misrepair-signatures denovo

DAMUTA signature definitions

• 18 Damage signatures
• 6 Misrepair signatures

nb. internally these signatures are referred to by their symbols in the graphical model: eta and phi respectively.

Model

Repo contents

• damuta python package
• docs package documentation
• example notebooks usage examples & tutorials
• manuscript code code to reproduce experiments and plots reported in manuscript

Documention

DAMUTA documentation is hosted via readthedocs

System requirements

Recommended Requirements:

• Unix-based system (tested on CentOS Linux 7)
• RAM: 16+ GB (for large-scale pan-cancer datasets)
• CPU: 4+ cores, 3.0+ GHz/core
• GPU: NVIDIA GPU with CUDA support (optional, for GPU acceleration via Theano)

Software dependencies are specified in (damuta_env.yml)[./damuta_env.yml]

Installation

From github

Clone this repo git clone https://github.com/morrislab/damuta

conda env create -f damuta_env.yml
conda activate damuta
pip install -e .

from PyPI

pip install damuta

theanorc

To use the GPU, ~/.theanorc should contain the following:

[global]
floatX = float64
device = cuda

Otherwise, device will default to CPU.

Installation Time Estimates

Conda environment setup: ~3-5 minutes Package installation: ~2-3 minutes

Demo

See quickstart to get started (~5 min)

Reproducing manuscript results

See manuscript code for code to reproduce experiments and plots reported in manuscript

Some data files are omitted from this repository due to access restrictions. Access can be requested from the corresponding sources:

• PCAWG/ICGC
• Hartwig
• Genomics England

Data for reproducing manuscript figures

Unrestricted-access data and certain useful intemediate files are also available via can be downloaded from zenodo

To download and organize these data:

# in top-level directory
wget  https://zenodo.org/records/15685052/files/damuta_zenodo.zip
unzip damuta_zenodo

mv damuta_zenodo/data/* manuscript/data
mv damuta_zenodo/figure_data/* manuscript/results/figure_data

# clean up now-empty directories
rmdir damuta_zenodo/data damuta_zenodo/figure_data damuta_zenodo

Some useful public data

file name	info	source
COSMIC_v3.2_SBS_GRCh37.csv	COSMIC database
icgc_sample_annotations_summary_table.txt	sample annotations used by PCAWG heterogeneity & evolution working group	ICGC data portal
PCAWG_sigProfiler_SBS_signatures_in_samples	counts of mutations attributed to each signature for PCAWG samples	syn11738669.7
pcawg_counts.csv	mutation type counts in PCAWG samples	Derived from syn7357330
pcawg_cancer_types.csv	sample annotations used in Jiao et. al	Adapted from z-scores file
gel_clinical_ann.csv	tumour type annotations for 18640 samples (ICGC, HMF, GEL)	Adapted from Degasperi et. al table S6
gel_counts.csv	mutation type counts for 18640 samples (ICGC, HMF, GEL)	Adapted from Degasperi et. al table S7

Citation

@misc{harrigan_damage_2025,
	title = {Damage and {Misrepair} {Signatures}: {Compact} {Representations} of {Pan}-cancer {Mutational} {Processes}},
	copyright = {© 2025, Posted by Cold Spring Harbor Laboratory. This pre-print is available under a Creative Commons License (Attribution-NonCommercial 4.0 International), CC BY-NC 4.0, as described at http://creativecommons.org/licenses/by-nc/4.0/},
	shorttitle = {Damage and {Misrepair} {Signatures}},
	url = {https://www.biorxiv.org/content/10.1101/2025.05.29.656360v1},
	doi = {10.1101/2025.05.29.656360},
	language = {en},
	urldate = {2025-06-02},
	publisher = {bioRxiv},
	author = {Harrigan, Caitlin F. and Campbell, Kieran and Morris, Quaid and Funnell, Tyler},
	month = jun,
	year = {2025},
}