Broad Institute

Pan-cancer clinico-genomic model for treatment response and survival prediction, transferring zero-shot to unseen hospitals and cancer types.

Histopathology foundation model predicting spatial gene expression from H&E slides at single-cell resolution via linear whole-slide attention.

Enzyme-substrate specificity prediction by end-to-end co-folding, with no predefined binding pocket. AUROC 0.766 on unseen enzymes and substrates.

Transformer that predicts protein-protein interactions at residue resolution, spanning mutations, PTMs, peptide-MHC binding, and disease variants.

Generative microscopy foundation model that synthesizes in-silico fluorescence images of protein subcellular localization from amino-acid sequence.

Long-context RNA foundation model that predicts splicing, isoform abundance, and variant effects from 64 kb of unspliced pre-mRNA sequence.

Prime editing efficiency prediction from pegRNA sequence, with every biochemical step of the editing mechanism modeled as its own learned rate.

Predicts single-cell scRNA-seq coverage and scATAC-seq insertion profiles from DNA sequence, adapting the Borzoi trunk with a cell-specific decoder.

Patient-level representation learning from scRNA-seq: a transformer set encoder with a diffusion decoder, fine-tuned for clinical prediction.

Multimodal foundation model predicting genome-wide binding of chromatin-associated proteins from protein sequence, DNA sequence, and chromatin state.

Splicing variant effect prediction across 49 human tissues and 15 developmental stages, from four weeks post conception to adulthood.

Single-cell perturbation-response model that predicts transcriptomic and imaging outcomes of unseen genetic perturbations via a VAE with attention.

Spatial proteomics foundation model, marker-aware and panel-agnostic, pretrained on 47 million multiplexed tissue-imaging patches from 175 markers.

De novo protein binder design that recasts structure-predictor confidence as an energy function, replacing ipTM as the hallucination objective.

Spatial transcriptomics prediction from H&E whole-slide images. One generative checkpoint covers 38,984 genes and 17 organs without fine-tuning.

Drug-combination safety prediction that fuses molecular structure, pathway knowledge, cell viability, and transcriptomic response to perturbation.

Genomic language model continue-pretrained on 13 million UK Biobank variants, giving variant-aware DNA embeddings for gene function and expression.

Slide-level pathology foundation model that encodes a whole-slide image of any size into one embedding, supervised by paired sequencing data.

Base-pair resolution sequence-to-activity CNN predicting ATAC-seq Tn5 insertion profiles and accessibility across 90 mouse immune cell types.

Latent diffusion model that paints high-resolution Cell Painting images of cells responding to a chemical compound or an over-expressed gene.

Histopathology patch encoder turning 512x512 tiles into 768-dimensional features, trained with a CoCa objective on 1.26 million captioned images.

Enhancer-promoter interaction prediction from DNA sequence and ATAC-seq alone. Spearman above 0.90 on cell types unseen during training.

Conditional autoregressive genomic language model trained on 13.6M mammalian promoters, scoring promoter variants, including indels, zero-shot.

De novo enzyme design conditioned on the reaction to be catalysed: substrate and product SMILES in, catalytic pocket, enzyme, and docked complex out.

Binding energy for protein-ligand, protein-protein, and antibody-antigen complexes is read off an energy model trained on crystal structures alone.

Enzyme optimum pH prediction from sequence, ensembling a light-attention network and a support vector regression over frozen ESM-1v embeddings.

Single-cell foundation model pretrained on about 30 million human transcriptomes, using rank-value encoding for context-aware gene network inference.