Combinatorial prediction of therapeutic perturbations using a causally-inspired neural network

Dataset supporting "Combinatorial prediction of therapeutic targets using a causally-inspired neural network"

Abstract: Phenotype-driven drug discovery, as an emerging alternative to target-driven strategies, focuses on identifying compounds that counteract the overall effects of diseases by analyzing phenotypic signatures. Our study introduces a novel approach to this field, aiming to expand the search space for discovering new therapeutic agents. We introduce PDGrapher, a causally-inspired graph neural network model designed to predict arbitrary perturbagens – a set of therapeutic targets – capable of reversing disease effects. Unlike current methods, which are limited by their reliance on predefined compound libraries, PDGrapher employs a novel combinatorial prediction framework to widen the search scope. PDGrapher has demonstrated significant improvements in predicting effective perturbagens, as evidenced by our evaluation across four datasets of genetic and chemical interventions. Notably, PDGrapher successfully predicted effective perturbagens in up to 10% additional test samples and ranked known therapeutic targets up to 35% higher than competing methods. A key innovation of PDGrapher is its direct prediction capability, which contrasts with the indirect, computationally-intensive models traditionally used in phenotype-driven drug discovery. This direct approach enables PDGrapher to train up to 30 times faster, representing a significant leap in efficiency and effectiveness. The results from our study suggest that PDGrapher could substantially advance phenotype-driven drug discovery, offering a faster, more expansive approach to identifying novel therapeutic compounds.