Learning Causal Gene Relationships in Biological Pathways with Graph Attention Networks (GATs)

Biological pathways are natural causal graphs mapping gene-gene interactions that govern human processes. Despite their importance, most ML models treat genes as unstructured tokens, discarding causal structure. The latest pathway-informed models capture pathway-pathway interactions, but still treat each pathway as a ``bag of genes" via MLPs, discarding its topology and gene-gene interactions. We propose a Graph Attention Network (GAT) framework that encodes gene-level pathway priors. We show that GATs generalize much better than MLPs, achieving an 81\% reduction in MSE when predicting pathway dynamics under unseen treatment conditions. We further validate the correctness of our biological prior by encoding drug mechanisms as causal graph modifications, improving robustness. Finally, trained without a prior, we show that our GAT model correctly rediscovers all five gene-gene interactions in the canonical TP53-MDM2-MDM4 feedback loop from raw time-series mRNA data, demonstrating the ability to learn causal gene relationships and generate novel biological insights directly from experimental data. [All code will be released upon publication.]