In a recent paper, University of Washington researchers argue that a key standard for deploying medical AI is transparency — that is, using various methods to clarify how a medical AI system arrives...

In a recent paper, University of Washington researchers argue that a key standard for deploying medical AI is transparency — that is, using various methods to clarify how a medical AI system arrives.....

Understanding why a model makes a certain prediction can be as crucial as the prediction's accuracy in many applications. However, the highest accuracy for large modern datasets is often achieved by c...

Large language models (LLMs) are having transformative impacts across a wide range of scientific fields, particularly in the biomedical sciences. Just as the goal of natural language processing is to ...

Allen School professor Su-In Lee, who directs the University of Washington’s AI for bioMedical Sciences (AIMS) Lab, is shaping the future of biology and medicine through artificial intelligence. Her r...

Allen School professor Su-In Lee, who directs the University of Washington’s AI for bioMedical Sciences (AIMS) Lab, is shaping the future of biology and medicine through artificial intelligence. Her r...

This issue highlights advances in applications of machine learning for diagnosing disease and for sorting and classifying health data, and includes a...

As the head of the Allen School’s Social Futures Lab, professor Amy X. Zhang’s research draws on the design of offline public institutions and communities to then develop new social computing systems ...

A key promise of sequence-to-function (S2F) models is their ability to evaluate arbitrary sequence inputs, providing a robust framework for understanding genotype-phenotype relationships. However, despite strong performance across genomic loci , S2F models struggle with inter-individual variation. Training a model to make genotype-dependent predictions at a single locus-an approach we call personal genome training-offers a potential solution. We introduce SAGE-net, a scalable framework and software package for training and evaluating S2F models using personal genomes. Leveraging its scalability, we conduct extensive experiments on model and training hyperparameters, demonstrating that training on personal genomes improves predictions for held-out individuals. However, the model achieves this by identifying predictive variants rather than learning a cis-regulatory grammar that generalizes across loci. This failure to generalize persists across a range of hyperparameter settings. These findings highlight the need for further exploration to unlock the full potential of S2F models in decoding the regulatory grammar of personal genomes. Scalable software and infrastructure development will be critical to this progress. ### Competing Interest Statement The authors have declared no competing interest.