A scalable approach to investigating sequence-to-function predictions from personal genomes.

SAGE-net framework for training and evaluating sequence-to-function models using personal genomes

SAGE-net is a scalable framework for training and evaluating sequence-to-function (S2F) models using personal genomes rather than reference genomes. S2F models trained on personal genomes partially improve inter-individual expression prediction, but gains remain modest in complex non-coding regions. The framework enables systematic benchmarking of existing S2F models on their ability to predict functional effects of personal variants. Code is open-source.

Synthesis written by Geno'X. For the full original abstract, please refer to the source publication.

Sequence-to-function models represent the next frontier for non-coding variant pathogenicity prediction—SAGE-net provides rigorous evaluation infrastructure that was missing in this domain. The modest gains reported are themselves valuable information for calibrating clinical expectations of these models.

Clinical impact: 2/3 · Evidence strength: 2/3 · Novelty: 1/2 · Sample size: 1/1 · Publication status: 1/1 → Total: 7/10