Enhancing Cross-Context Generalization in Drug Perturbation Prediction with a Multimodal Conditional Diffusion Framework.

Abstract

Predicting drug-induced transcriptional perturbations is critical for precision medicine, yet existing models fail to capture multimodal biological context, limiting generalization across unseen drugs and cell lines. We present PertDiff, a conditional diffusion framework that integrates control gene expression, LLM-derived cell semantics, and pretrained molecular graph representations to predict transcriptome-wide perturbations. PertDiff outperforms state-of-the-art baselines in prediction accuracy and generalizes robustly across drugs and cell lines. It further demonstrates translational utility through accurate drug sensitivity prediction, therapeutic repurposing for pancreatic cancer, and concordance with real-world clinical treatment outcomes, establishing it as a biologically grounded transcriptomic modeling tool. The source code and data are available at https://github.com/Panda-myj/PertDiff and https://doi.org/10.5281/zenodo.18427848. Supplementary data are available at Bioinformatics online.