Abstract 5813: AI-driven identification and validation of novel synthetic lethal gene pairs through deep mining of cancer dependency data

Abstract

Abstract The targeting of synthetic lethal (SL) gene pairs is emerging as a new paradigm in precision oncology, with the potential to deliver treatments with higher efficacy and fewer side-effects. However, existing approaches have focused on a few well-validated pairs, and there is a clear need to expand the toolbox of potential SL targets. We have performed deep mining of cancer dependency data using a suite of machine learning tools and AI algorithms underpinned by robust statistical analysis to identify the next generation of SL gene pairs. To ensure robust, high confidence validation of multiple SL pairs simultaneously, we have used three cell lines and orthogonal assay formats. This has led to the confirmation of new druggable SL genes suitable for clinical development. We have identified the cancer subtypes with the strongest evidence of SL vulnerabilities and quantified patient populations who would most benefit from novel therapeutic agents. For one gene pair, we have generated novel best-in-class inhibitors of the target of interest, with exceptional selectivity over the SL partner, and used these chemical tools to validate the SL relationship in 2D and 3D models. Citation Format: Daniel S. Miller, Oliver Vipond, Alfie Brennan, Anna Hercot. AI-driven identification and validation of novel synthetic lethal gene pairs through deep mining of cancer dependency data [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 5813.