Variational and Explanatory Neural Networks for Encoding Cancer Profiles and Predicting Drug Responses

• DOI: 10.48550/arxiv.2407.04486
• Publication Date: 2024-07-05
• Authors: Tianshu Feng; Rohan Gnanaolivu; Abolfazl Safikhani; Yuanhang Liu; Jun Jiang; Nicholas Chia; Alexander Partin; Priyanka Vasantha...; Yitan Zhu; Chen Wang
• Keywords: Explanatory model

Human cancers present a significant public health challenge and require the discovery of novel drugs through translational research. Transcriptomics profiling data that describes molecular activities in tumors cancer cell lines are widely utilized for predicting anti-cancer drug responses. However, existing AI models face challenges due to noise transcriptomics lack biological interpretability. To overcome these limitations, we introduce VETE (Variational Explanatory Encoder), neural network framework incorporates variational component mitigate effects integrates traceable gene ontology into architecture encoding data. Key innovations include local interpretability-guided method identifying paths, visualization tool elucidate mechanisms responses, application centralized large scale hyperparameter optimization. demonstrated robust accuracy line classification response prediction. Additionally, it provided explanations both tasks offers insights underlying its predictions. bridges gap between AI-driven predictions biologically meaningful research, which represents promising advancement field.