Princeton University

Pan-cancer pretrained diffusion model imputing genome-wide expression from sparse spatial transcriptomics panels, zero-shot and reference-free.

Splice donor and acceptor site prediction from raw DNA, scoring every position of a 20 kb window with an ensemble of dilated residual CNNs.

Isoform-resolved variant effect prediction from DNA sequence, using graph attention over transcript splice structures across 30 human tissues.

Antibody CDR design model fusing diffusion with a GFlowNet objective so binding energy is optimized during training rather than by post-hoc RL.

Spike inference from calcium imaging traces, driven by a multistate GCaMP kinetic model that generates the synthetic data its decoders are trained on.

RNA foundation model unifying sequence representation, 3D structure prediction, and de novo design. Ranks first on 11 of 13 BEACON tasks.

Cryo-EM heterogeneous reconstruction that models particles as one of K neural fields, resolving compositional and conformational states ab initio.

Spatial gene expression prediction from H&E tumor histology, aligning a pathology foundation model with a single-cell RNA-seq foundation model.

Chemical language model of the human metabolome that generates and ranks candidate structures for unidentified mass spectrometry peaks.

Neural ab initio reconstruction for cryo-EM and cryo-ET that jointly infers particle poses and a continuous landscape of conformational states.

Transformer language model for 5' UTR sequences that predicts mRNA translation efficiency, ribosome loading, and protein expression levels.

fMRI foundation model pretrained with masked autoencoding on roughly 6,700 hours of recordings for clinical prediction and network discovery.

Chromatin-level variant effect prediction from DNA sequence, projecting 21,907 predicted regulatory profiles onto 40 interpretable sequence classes.

RNA-binding protein target site prediction from 1,000 bp of sequence, scoring how a noncoding variant disrupts binding across 88 RBPs.

Tissue-specific gene expression prediction from DNA sequence, scoring a noncoding variant as the log fold change it causes in each of 218 tissues.

Chromatin feature prediction from 2 kb of DNA, scoring 2,002 transcription factor, DNase and histone profiles to rank noncoding variant effects.

Noncoding variant effect prediction from DNA sequence, scoring how an allele shifts 919 chromatin features across ENCODE and Roadmap cell types.