Biologically Informed Deep Neural Networks for Multi-Omic Integration, Pathway Activity Inference and Risk Stratification in Cancer
2026-07-06 • Machine Learning
Machine Learning
AI summaryⓘ
The authors developed a new method called Pathway Activity Autoencoders to combine different biological data types in a way that balances understanding what the model learns with capturing complex details. They tested their approach on breast cancer data to predict patient outcomes and cancer subtypes, finding that combining gene, protein, and microRNA data worked best. Their method also showed consistent performance when used carefully without too much regularization. Visualizations helped show the model's results make biological and clinical sense. Overall, the authors provide a useful tool for integrating biological data with clear connections to disease insights.
multi-omicsautoencoderpathway analysisbreast cancersurvival predictionsubtype classificationgene expressionprotein expressionmicroRNAregularization
Authors
Pedro Henrique da Costa Avelar, Le Ou-Yang, Min Wu, Sophia Tsoka
Abstract
Integrating complex, multi-omics data presents significant challenges. Existing approaches often face a trade-off between model interpretability and representational capacity, with most either relying on post-hoc interpretation or use linear models that may overlook complex interactions. We report Pathway Activity Autoencoders for the multi-omics setting, which embed prior knowledge via pathway-informed architectural constraints, fostering interpretability, while preserving representational power. Our multi-omic framework is applied in the context of breast cancer and is evaluated in survival prediction and subtype classification with results indicating a positive effect of integration. We conduct analysis of individual omics layer impact on end-task performance, revealing that gene, protein, and microRNA expression layers provide the strongest contribution. Repeatability studies indicate that, while dropout improves model robustness and consistency, excessive regularisation can reduce predictive performance. Finally, visualizations of the learned feature space illustrate the framework's intrinsic transparency and clinical relevance. The results underscore the value of multi-omic integration and delineate the impact of individual omics layers, establishing practical guidelines for integration within our framework. Overall, our pathway activity autoencoder frameworks yield superior latent representations that are biologically meaningful and are directly translatable into clinically relevant insights.