Biologically informed neural network models are robust to spurious interactions via self-pruning

Computational models of cellular networks hold promise to uncover disease mechanisms and guide therapeutic strategies. Biology-informed neural networks (BINNs) is an emerging approach to create such models by combining the predictive power of deep learning with prior knowledge, a vital aspect of biological research. The architectures of BINN's enforces a network structure from which mechanism can ideally be inferred. However, a key challenge is to evaluate the reliability of these mechanisms, as cells are inherently complex, involving intricate and sometimes unknown interactions. Currently, analysis mainly focuses on selected pathways rather than a more comprehensive perspective. In this work we demonstrate an improved, holistic approach: we measure to which extent purposefully introduced spurious interactions are removed by a BINN during training (self-pruning). This metric is scalable and generalizable, as it does not depend on manual curation and can be translated into diverse network settings. To enable rapid testing, we reimplemented LEMBAS (Large-scale knowledge-EMBedded Artificial Signaling-networks), our recurrent neural network framework for intracellular signaling dynamics, with full GPU acceleration. Our implementation achieves a >7-fold speedup compared to the original while preserving predictive accuracy. We evaluated self-pruning in 3 different datasets and found that when spurious interactions are introduced at random, the model prunes these to a larger extent than those from the prior knowledge network (PKN), provided the model is regularized with a sufficiently large L2 norm. This suggests that BINNs are robust to uncertainty in the PKN and is a quantitative sign that they could model real aspects of the systems. Our implementation of LEMBAS is freely available under a MIT license at https: //github. com/AvlantNilssonLab/LEMBASGPU. The models and results to generate the figures can be downloaded through https: //zenodo. org/records/17425598.