# Regulators of homologous recombination deficiency identified by machine learning using somatic multi-omics data

## TL;DR

This study uses machine learning and multi-omics data to discover new genetic factors linked to homologous recombination deficiency in various cancers.

## Contribution

The study introduces a machine learning framework that identifies novel regulators of homologous recombination deficiency beyond BRCA1/2.

## Key findings

- The model achieved high predictive performance using somatic multi-omics data from over 8,000 patients.
- SHAP-based analysis revealed shared and cancer-specific molecular determinants of HRD.
- Findings expand the known HRD-associated alterations and suggest integrative AI can improve patient stratification for HR-targeted therapies.

## Abstract

Using somatic multi-omics data and explainable artificial intelligence, this study identifies novel alterations underlying homologous recombination repair deficiency across cancers.

Homologous recombination deficiency (HRD) is a critical biomarker for guiding targeted therapies, yet the full range of somatic alterations driving HRD across cancers remains incompletely characterized. Here, we present a tumor-agnostic machine learning framework that integrates somatic multi-omics data, including copy-number variations, single-nucleotide variants, DNA methylation, and gene expression from over 8,000 patients in The Cancer Genome Atlas. Using a genome-wide mutational signature–based HRD score as ground truth, our model achieved high predictive performance and leveraged SHAP-based explainability to uncover HRD regulators beyond BRCA1/2. Cross-tumor analysis revealed both shared and cancer type–specific molecular determinants, whereas functional enrichment highlighted key molecular and cellular processes. These findings expand the known repertoire of HRD-associated alterations, provide a resource for mechanistic investigation, and demonstrate the potential of integrative AI approaches to improve patient stratification for HR-targeted therapies across diverse malignancies.