PANTHER: Pathway Augmented Nonnegative Tensor Factorization for HighER-order Feature Learning

Genetic pathways usually encode molecular mechanisms that can inform targeted interventions. It is often challenging for existing machine learning approaches to jointly model genetic (higher-order features) and variants (atomic features), present clinicians interpretable models. In order build more accurate better models medicine, we introduce Pathway Augmented Nonnegative Tensor factorization HighER-order feature (PANTHER). PANTHER selects informative directly mechanisms. We apply genetically motivated constrained tensor group in a way reflects mechanism interactions. then train softmax classifier disease types using the identified pathway groups. evaluated against multiple state-of-the-art tensor/matrix models, as well guided Bayesian hierarchical outperforms all comparison significantly (p