Gene dependency-informed inference of response to targeted cancer therapies

2026-06-08 · Source: Machine learning : nature.com subject feeds · Field: Health & Wellbeing — Pharmaceuticals & Biotechnology, Medical Specialties & Subspecialties, Health & Medical Research · Depth: Expert, short

Summary

FORGE (Factorization Of Response and Gene Essentiality) is a novel joint matrix factorization framework designed to co-model drug response and target gene dependency, providing biologically informed stratification for targeted cancer therapies. This framework addresses the lack of mechanistic grounding in existing omics-based drug sensitivity models. FORGE generates a "Benefit Score" from basal gene expression to predict therapeutic potential. In tests with unseen cell lines treated with erlotinib, FORGE demonstrated high concordance for gene dependency (0.69) and IC50 (0.62). Stratification by Benefit Score showed increased dependency and decreased IC50 across quartiles. The joint modeling approach improved predictive performance over single-task methods and enhanced agreement between gene-level effects (p = 0.039). Validation across independent datasets, including patient-derived xenografts and the Tahoe-100M dataset, confirmed that higher Benefit Scores correlate with tumor regression and predicted drug susceptibility. Furthermore, FORGE identifies underlying gene programs linked to drug susceptibility.

Key takeaway

For research scientists developing precision oncology solutions, you should consider integrating gene dependency data with drug response modeling. FORGE's joint matrix factorization approach offers a robust method to predict therapeutic potential and stratify patient groups more effectively. This can lead to improved mechanistic understanding of drug susceptibility and more accurate predictions for targeted therapies, guiding your experimental design and clinical translation efforts.

Key insights

FORGE co-models drug response and gene dependency to predict cancer therapy benefit, improving mechanistic understanding and prediction.

Principles

• Joint modeling enhances predictive accuracy.
• Gene dependency informs drug sensitivity.
• Basal gene expression predicts therapeutic potential.

Method

FORGE uses joint matrix factorization to co-model drug response and target gene dependency, deriving a Benefit Score from basal gene expression for stratification.

In practice

• Stratify cancer patients for targeted therapies.
• Identify gene programs for drug susceptibility.
• Improve drug sensitivity prediction models.

Topics

• Targeted Cancer Therapies
• Gene Dependency
• Drug Response Prediction
• Matrix Factorization
• Precision Oncology
• Omics-based Modeling

Best for: AI Scientist, Research Scientist, Data Scientist

Related on AIssential

• Predicting and interpreting cell-type-specific drug responses in the small-data regime using inductive priors — PrePR-CT uses graph-based deep learning with cell-type-specific co-expression networks to predict drug responses in data-limited settings.
• AI Improving Dose Decisions and Patient Outcomes in Oncology- with Shefali Kakar of Novartis — AI-driven data integration refines oncology dose decisions, safety, and patient variability, accelerating drug development.
• Meta-encoder: a unified integration framework for multiple pathological foundation models in cancer detection — The Meta-Encoder integrates multiple pathological foundation models to improve cancer detection and molecular characterization.
• Critical evaluation of drug response prediction models with DrEval — Current drug response models often yield overly optimistic performance due to evaluation biases and poor generalization.
• Uncovering repurposed medicines to fight liver fibrosis — AI-driven drug repurposing, specifically using Co-Scientist, effectively identifies novel anti-fibrotic drug candidates, outperforming human selection in initial tests.
• Probe Before You Edit: Probing-Guided Molecular Optimization for LLM Agents in Structure-Based Drug Design — Probing molecular edit responses guides LLM agents to jointly optimize binding affinity and druggability in drug design.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Machine learning : nature.com subject feeds.