All Competitors

Matryoshka sparse autoencoders scaled to ESM2-3B, exposing the features that drive protein structure prediction and steering ESMFold outputs.

Protein model accuracy estimation predicting per-residue lDDT plus signed residue-pair distance errors that become Rosetta refinement restraints.

Diffusion-based backbone generation and sequence design method for programmable asymmetric transmembrane beta-barrel nanopores.

Structure-based drug design model generating 3D ligands in a protein pocket with interaction-guided flow matching and a learned atom-count predictor.

De novo protein backbone generator built on rectified quaternion flow matching, reaching 0.972 designability with far fewer sampling steps.

Structure-based drug design model that generates 3D ligands inside a protein pocket by interpolating distribution parameters instead of samples.

Multimodal protein representation model that iteratively fuses a sequence language model with a 3D structure encoder through a shared learnable token.

De novo protein backbone design conditioned on a target per-residue flexibility profile, with SE(3)-equivariant flow matching and MD validation.

Hallucination framework for de novo nucleic acid design, pairing NA-MPNN sequence proposals with a frozen AlphaFold3 or Protenix structure oracle.

De novo protein backbone generation and sequence-conditioned folding using SE(3) flow matching over a physics-based, clash-free unfolding process.

Protein complex model quality assessment via DockQ-guided graph contrastive learning. CASP16 TMscore ranking loss of 0.123 versus 0.138 runner-up.

Latent diffusion model for controllable all-atom protein generation that co-designs sequence and structure while training on sequences alone.

Diffusion model that backmaps coarse-grained protein structures to all-atom detail, scaling to condensates of over a million residues.

Distilled few-step protein backbone generator that adapts Score Identity Distillation to Proteina for over 20x faster de novo structure sampling.

Unified all-atom generative model for biomolecular structure prediction, binder filtering, and controllable protein and nanobody design.

Multi-domain protein and complex assembly from deep-learned inter-domain interactions, averaging TM-score 0.922 across 219 multi-domain targets.

Protein language model trained on roughly 2.8 billion sequences, forming the representation core of Biohub's world model of protein biology.

Latent diffusion model that designs D-peptide binders against native L-protein targets, generalizing across chirality via axial vector features.

Pfam domain annotation with a graph neural network over AlphaFold2 residue-interaction graphs and ESM-2 embeddings, one model for all families.

Structure-free peptide binder design conditioned only on a target protein sequence, using contrastive alignment to steer a latent diffusion model.

Genomic foundation model trained on 9.3 trillion DNA base pairs across all domains of life, with 40B parameters and a 1-million-token context.

Transformer protein language model trained on 250 million protein sequences that learns structural and functional representations without supervision.

Retrieval-augmented protein language model that conditions ESM-2 on homologous sequences via cross-attention for conditional sequence generation.