All Competitors

RNA-protein complex refinement via diffusion, repositioning the protein against the RNA to improve AlphaFold 3 and ProRNA3D-single backbones.

Inverse protein folding model for all-atom structures with bound ligands, nucleotides, or metal ions. Reaches 75.7% sequence recovery at metal sites.

Protein-protein interaction predictor that adds contact-guided dual attention and a geometric encoder to frozen protein language model embeddings.

Allosteric binding site prediction from protein sequence alone, using LoRA-tuned protein language models conditioned on the orthosteric pocket.

Automated cryo-EM structure determination that fuses density maps with AlphaFold3 predictions, averaging a TM-score of 0.93 on high-resolution maps.

All-atom biomolecular structure prediction with adapters for allosteric states, user-defined interface constraints, and binding affinity.

Backmapping model that rebuilds all-atom protein and nucleic acid structures from coarse-grained beads and inpaints unresolved residues.

Proteomics foundation model for peptide-spectrum scoring and open de novo sequencing, reading over 1,300 modifications from tandem mass spectra.

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

Allele-free HLA class I epitope classification from peptide sequence alone, via LoRA-adapted ESM-2 with parallel CNN and Transformer branches.

Conditional GAN generating HLA class I pseudo-sequences from a peptide, then resolving them to candidate alleles by nearest-neighbor lookup.

T-cell receptor specificity prediction that separates general antigens from autoimmune-related ones using ESM-2 embeddings and a topology-aware graph.

CATH superfamily classifier over ProstT5 amino-acid and 3Di structural-alphabet embeddings, reaching 92.2% accuracy on roughly 1,700 superfamilies.

Antibody sequence-structure co-design diffusion model adding atom-level equivariant geometry to residue embeddings, raising CDR-H3 recovery to 38.9%.

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

De novo peptide sequencing from tandem mass spectra with a non-autoregressive Transformer trained on 100 million peptide-spectrum matches.

Single-cell diaPASEF proteomics search that scores coelution with a pretrained CNN and returns a protein matrix with no missing values.

Protein backbone generation from voxelized 3D densities, pairing a 3D CNN autoencoder with latent flow matching for shape and motif conditioning.

Protein inter-residue distance prediction fusing MSA Transformer coevolution with ESM2 sequence features, reaching a mean absolute error of 2.20 Å.

Catalytic residue prediction from enzyme sequence alone, pairing frozen ESM2 embeddings with contrastive learning. Reaches F1 0.86 on Uni3175.

Structure-free ligand generation conditioned on a protein sequence alone, using masked diffusion over SMILES trained on 1.2M BindingDB active pairs.

Peptide-HLA immunogenicity prediction with a BiLSTM ensemble, inside a pipeline that finds microbial epitopes mimicking tumor neoantigens.

Structure-based drug design model pairing SE(3)-equivariant diffusion with retrieval of pocket-matched scaffolds to generate ligands for a target.