{"entity": "researcher", "timestamp": "2026-08-18T14:16:11.237Z", "family": "Xu", "given": "Xuechun", "initials": "X", "orcid": "0000-0003-1850-0946", "affiliations": ["Department of Cell and Molecular Biology, Karolinska Institutet, Stockholm, 17177, Sweden.", "Science for Life Laboratory (SciLifeLab), Karolinska Institutet, Stockholm, 17177, Sweden."], "links": {"self": {"href": "https://publications-affiliated.scilifelab.se/researcher/2d02b83768c24f5396b85938a9590b7f.json"}, "display": {"href": "https://publications-affiliated.scilifelab.se/researcher/2d02b83768c24f5396b85938a9590b7f"}}, "publications": [{"entity": "publication", "iuid": "b18a8d3139574233b6caaae6a511053a", "links": {"self": {"href": "https://publications-affiliated.scilifelab.se/publication/b18a8d3139574233b6caaae6a511053a.json"}, "display": {"href": "https://publications-affiliated.scilifelab.se/publication/b18a8d3139574233b6caaae6a511053a"}}, "title": "Biologically informed neural network models are robust to spurious interactions via self-pruning.", "authors": [{"family": "Nordenstorm", "given": "Olof", "initials": "O"}, {"family": "Baghdassarian", "given": "Hratch", "initials": "H"}, {"family": "Xu", "given": "Xuechun", "initials": "X", "orcid": "0000-0003-1850-0946", "researcher": {"href": "https://publications-affiliated.scilifelab.se/researcher/2d02b83768c24f5396b85938a9590b7f.json"}}, {"family": "Lauffenburger", "given": "Douglas A", "initials": "DA", "orcid": "0000-0002-0050-989X", "researcher": {"href": "https://publications-affiliated.scilifelab.se/researcher/ac0a09cb7c9c49de95b660756a4e5464.json"}}, {"family": "Nilsson", "given": "Avlant", "initials": "A", "orcid": "0000-0002-9476-4516", "researcher": {"href": "https://publications-affiliated.scilifelab.se/researcher/f2e21dbc1c624f6a841c59e959e948e4.json"}}], "type": "journal article", "published": "2026-08-03", "journal": {"title": "Bioinformatics", "issn": "1367-4811", "volume": "42", "issue": "8", "issn-l": "1367-4803"}, "abstract": "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 models, as cells are inherently complex, involving intricate and sometimes unknown interactions. Currently, analysis mainly focuses on selected pathways rather than a more comprehensive perspective.\n\nIn this work we demonstrate an alternative holistic approach: we measure to which extent purposefully introduced spurious interactions are down-weighted by a BINN during training (self-pruning). The metric suggested RRA (Relative Residual Area) allows for direct distribution comparison with perfect self-pruning achieved at zero and a failure to self-prune if above one. To enable rapid testing, we updated 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 can be robust to uncertainty in the PKN.\n\nOur implementation of LEMBAS is freely available under a MIT license at https://github.com/AvlantNilssonLab/LEMBAS_GPU. The models and results to generate the figures can be downloaded through https://zenodo.org/records/17425598.", "doi": "10.1093/bioinformatics/btag556", "pmid": "42502984", "labels": {"Avlant Nilsson": null, "DDLS Fellow": null}, "xrefs": [{"db": "pmc", "key": "PMC13457084"}, {"db": "pii", "key": "8742075"}], "notes": [], "created": "2026-08-17T11:11:10.279Z", "modified": "2026-08-17T11:13:01.812Z"}]}