All Competitors

Protein-ligand scoring function that conditions probabilistic geometric potentials on language model priors to rank docked poses and binding affinity.

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

E(3)-equivariant diffusion model for macrocycle design that turns acyclic molecules into macrocycles, with a transformer choosing where to cyclize.

Protein structure tokenizer that encodes a whole structure globally, with each successive token adding detail for adaptive-length representations.

Protein backbone generator running a diffusion transformer over SaProt structural tokens, with an IPA token cache to speed up de novo design.

Single-cell perturbation prediction model using conditional flow matching to map control cells to perturbed expression distributions.

Protein structural homology search from sequence alone, embedding proteins so that structural similarity becomes a fast nearest-neighbor lookup.

Protein language model that predicts per-residue local energetic frustration directly from sequence, covering whole proteomes and disordered regions.

Connects a frozen protein language model to a general LLM via a cross-modal projector, adding protein reasoning without catastrophic forgetting.

Masked language model for T-cell receptor and peptide-MHC binding prediction, with compositional pretraining and non-autoregressive decoding.

mRNA language foundation model trained on ~115M protein-coding sequences across the tree of life, unifying mRNA perception and generation.

Multimodal model that designs small molecules from transcriptomic and cell-imaging perturbation phenotypes with a rectified flow transformer.

Single-cell foundation model built on masked discrete diffusion, jointly generating gene identities and expression values from 59 million cells.

Genome language model that adds evolutionary-rate prediction to pretraining, improving representations for variant effect and regulatory genomics.

Vision-language DNA model that renders genomic sequence as visual layouts, reading regions up to 450,000 bases with about 20x better token efficiency.

Causal 309M-parameter protein language model that scores variant fitness zero-shot and generates sequences, reaching 0.390 Spearman on ProteinGym.

Family of autoregressive genomic foundation models that reconcile k-mer tokenization with single-nucleotide resolution at contexts up to 98k bp.

Retrieval-augmented genomic foundation model that gives transformer backbones a hash-based k-mer motif memory for functional genomics tasks.

Large causal cell model trained on cancer perturbation data that generalizes zero-shot to patient-derived cells for therapeutic target prioritization.

Enzyme design model that jointly generates enzyme sequences and substrate-binding pockets, conditioned on functional priors and substrate structure.

Partially latent flow-matching model for de novo protein design, jointly generating sequence and all-atom structure for proteins up to 800 residues.

Flow-matching generative model for de novo protein binder backbone design, built on a Pairformer architecture with in silico interface maturation.

LoRA adapter on ProstT5 predicting per-residue distributions over Foldseek 3Di tokens, capturing conformational flexibility from MD trajectories.

Structure-free protein-ligand binding affinity predictor built on OpenFold3 that scores potency from a protein sequence and a ligand SMILES string.