Stanford University

Pocket-conditioned 3D ligand generator that steers flow matching with LLM-written chemical priors, for de novo design and scaffold hopping.

Sequence-conditioned generative framework that reconstructs missing prostate MRI contrasts and restores artefact-degraded acquisitions.

Vision-language model for neuroblastoma pathology that reads H&E slides with their reports to grade tumors, infer biomarkers and stratify risk.

Generative framework that learns a developmental vector field from scRNA-seq snapshots, coupling flow matching with molecular RNA kinetics.

Graph foundation model for fMRI brain networks, pretrained across 27 datasets with graph and language prompts for zero-shot disorder classification.

Multimodal Q-former that fuses DNA sequence, gene context, protein function, and text for zero-shot variant interpretation with a frozen LLM.

Reasoning-guided foundation model for de novo antibody CDR design, pairing a multimodal LLM understanding expert with a Boltz-1 diffusion expert.

110M-parameter RNA language model that designs sequences from secondary structure, motif, and Gene Ontology constraints via discrete diffusion.

Generative pipeline for epitope-targeted de novo antibody (nanobody) CDR design that yields nanomolar binders from only dozens of designs per antigen.

Spatial transcriptomics foundation model pairing gene expression with H&E histology for spatial domain discovery and clinical outcome prediction.

Diffusion-transformer pathology model embedding H&E histology, RNA profiles, and clinical text in a latent space for zero-shot cross-modal synthesis.

Graph attention model that learns context-aware protein embeddings from protein-protein interaction, co-expression, and tissue association networks.

Protein-ligand binding affinity scorer using an SE(3)-equivariant graph network trained on 741,706 co-folded complexes with target-disjoint splits.

Masked DNA language model for regulatory genomics with a motif-discovery regularizer for zero-shot TF motif recovery and variant effect prediction.

Graph-attention model that predicts A-to-I RNA editing from sequence and secondary structure, treating RNA as a graph with base-pairing edges.

Spatial transcriptomics foundation model built on a lightweight graph convolutional network and trained by masked central-spot prediction.

Single-cell foundation model using tabular attention over context cells to predict responses to arbitrary perturbations without fine-tuning.

Flow-matching model for therapeutic peptide design that co-designs sequence, structure, and molecular surface to disrupt protein-protein interactions.

3D vision-language foundation model for abdominal CT, pretrained on scans, radiology reports, and EHR codes for zero-shot interpretation.

Tissue imaging foundation model pretrained on matched H&E histology and spatial proteomics for cross-modal inference and zero-shot retrieval.

Hierarchical single-cell foundation model that turns scRNA-seq profiles into zero-shot donor-level embeddings for disease and biomarker prediction.

Denoising diffusion bridge model for peptide binder design that generates ligand surfaces and backbones complementary to a target receptor surface.

All-atom protein representation model that learns from each residue's strictly local atomic neighborhood, capturing side-chain geometry and chemistry.

Potts-model inverse folding that conditions on a structural ensemble rather than a single backbone, improving designability and self-consistency.

De novo antibiotic design framework coupling a 6.4B-parameter protein language model with reinforcement learning to generate antimicrobial peptides.

Micronuclei instance segmentation for fluorescence microscopy, using an anchor-tuned Mask R-CNN to detect micronuclei and link them to parent nuclei.

DNA methylation foundation model that learns blood aging as a continuous ODE and prescribes sparse CpG edits for in-silico rejuvenation.

All-atom protein structure diffusion models for motif scaffolding and hotspot-conditioned complex generation, at 22M parameters.

Retinal OCT vision-language model that writes layer-by-layer clinical summaries and assigns six-class disease labels from a single B-scan.

Peptide toxicity prediction that fuses frozen ProtT5 residue embeddings with ESMFold-predicted structure in an E(3)-equivariant graph neural network.

Multi-modal protein language model using the MSA evolutionary profile as a reasoning step between structure and sequence. 650M outperforms ESM-3 1.4B.

Mutational effect predictor for protein-protein binding energy, matching the FoldX force field's accuracy with a 1,000x speed-up.

Conditional diffusion model for 7T brain MRI denoising that turns a single 5-minute gradient-echo scan into a four-repetition-quality image.

Self-supervised medical imaging foundation model pretrained on 3.3 million CT, X-ray, ultrasound, pathology, OCT, fundus, and dermoscopy images.

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

Structure-based drug design by diffusing medicinal-chemistry fragments into a binding pocket, yielding synthesizable, selective, drug-like molecules.

General medical vision-language model trained with reinforcement learning to reason step by step over medical images for diagnosis and visual QA.

Generative diffusion foundation model for human gait dynamics that estimates ground reaction forces from partial motion-capture kinematics.

Clinical imaging encoder multitask-pretrained across X-ray, mammography, dermoscopy, fundus, ultrasound, CT, and histopathology for few-shot transfer.

Predicts haplotype-specific 3D genome organization and Hi-C contact maps from a single long-read Fiber-seq assay, using no DNA sequence as input.

Fixed-backbone protein sequence design that co-generates amino acid identity and sidechain conformation, with 49.7% sequence recovery on CATH 4.2.

Prostate cancer detection model for MRI and transrectal ultrasound, trained with patch-level contrastive learning across 4,401 patients.

Direct RNA nanopore basecaller that translates raw ionic-current signal into nucleotide sequence using a hybrid Mamba-Transformer backbone.

Breast ultrasound generative foundation model that synthesizes conditioned images to train screening, diagnosis, and prognosis models.

Vision-language foundation model for precision oncology, pretrained on 50M pathology images and 1B text tokens via unified masked modeling.

Single-cell foundation model pretrained by federated learning, modeling expression as a cell-by-gene table rather than an ordered gene sentence.

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

Base-resolution chromatin accessibility model that factors out enzyme sequence bias to score regulatory variants and transcription factor footprints.

Cell type annotation from multiplexed tissue images, using a pretrained Vision Transformer ensemble that runs on new panels without fine-tuning.

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

Vision transformers trained on Human Protein Atlas fluorescence microscopy for subcellular protein localization and cell morphology representation.

Whole-slide pathology assistant that states the morphological findings behind each diagnosis, trained on 180k VQA pairs from 9,850 gigapixel slides.

Histopathology foundation model aligned to spatial transcriptomics by a cross-modal ranking loss, embedding H&E patches without gene input.

Patient-level foundation model that pools every cell in an scRNA-seq sample into one disease representation, trained on 24.3 million cells.

Sparse autoencoders on ESM-2 embeddings that expose thousands of interpretable features per layer, tied to binding sites, motifs, and domains.

Virtual multiplex immunofluorescence staining from H&E histopathology, imputing the expression and spatial localization of 50 protein biomarkers.

Contrastive dual-encoder aligning T-cell receptor CDR3 and peptide epitope sequences in one latent space to rank which receptors bind which antigens.

Vision-language assistant that reads a whole gigapixel pathology slide, answering diagnostic questions and writing slide-level descriptions.

Protein sequence embedding method that pools a language model's token outputs by PageRank over its own attention, adding no trained parameters.

Cis-regulatory element classifier that reads DNA sequence plus chromatin accessibility and loop tracks to label enhancers, silencers and insulators.

Weakly supervised histopathology foundation model pretrained on 60,530 whole-slide images for cancer detection, prognosis, and molecular prediction.

Open-source, lightweight generalist vision-language foundation model for diverse biomedical imaging and text tasks.

Medical vision-language model trained on the MedTrinity-25M dataset, answering questions and generating text about radiology and histology images.

Semi-supervised cryo-ET segmentation framework that adapts DINOv2 vision transformers for 3D organelle annotation using sparse 2D slice labels.

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

Multi-modal foundation model for sleep analysis, learning joint representations across brain, cardiac, and respiratory polysomnography signals.

Structure-conditioned protein language model aligned to experimental stability data, scoring variant stability and generating stabilized sequences.

Self-supervised 3D CT foundation model that extracts general-purpose tumor representations for cancer imaging biomarker discovery and prognosis.

RNA foundation model trained on chemical-mapping data from millions of sequences, predicting reactivity, secondary structure, and degradation.

Instruction-tuned vision-language foundation model for chest X-ray interpretation, with 8 billion parameters spanning eight clinical task types.

Single-cell foundation model producing species-agnostic cell embeddings by representing genes through frozen ESM-2 protein language model embeddings.

Pseudo-membrane generator for fluorescence microscopy that synthesizes missing membrane staining from nuclei to improve single-cell segmentation.

Discrete diffusion model for protein sequence and MSA generation, enabling controllable de novo design directly in sequence space without structure.

Vision-language foundation model for pathology, fine-tuned from CLIP on 208,414 image-text pairs for zero-shot classification and image retrieval.

Multimodal medical vision-language model for few-shot visual question answering, learning new imaging tasks from in-context examples at inference.

Genomic foundation model built on the Hyena operator, processing DNA at single-nucleotide resolution with context windows up to 1 million tokens.

Spatial proteomics imputation from a 7-plex immunofluorescence panel, generating in silico CODEX expression for 33 more biomarkers per cell.

Zero-shot antibody affinity maturation using ESM pseudolikelihood scoring. Improves binding up to 160-fold with no antigen-specific training data.

Text-conditioned latent diffusion model that generates synthetic chest X-rays from free-form radiology prompts by adapting Stable Diffusion.

Self-supervised vision-language model for zero-shot detection of chest X-ray pathologies, trained on image-report pairs without explicit labels.

Protein structure and complex prediction from sequence, in a three-track network that reasons over alignments, distances, and 3D coordinates at once.