All Competitors

Chest radiograph and report encoder pretrained contrastively on a million pairs, whose frozen embeddings train classifiers from a few hundred cases.

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

Discrete flow generative model over fragmented SMILES for de novo, fragment-constrained, and property-optimized small-molecule drug design.

Scientific multitask language model whose byte-level genome modeling beats Evo 7B on DNA perplexity and mutation-effect prediction at 1.5B params.

Sequence-based protein-ligand binding site predictor pairing a protein language model with a SMILES chemical language model for zero-shot ligands.

Protein structure prediction model that weights residues by protein-language-model importance scores to improve accuracy on hard AlphaFold2 targets.

Multi-label classifier over ESM-2 embeddings that separates DNA-binding, RNA-binding, and dual-binding proteins using label-aware attention.

DNA sequence embedding model that approximates edit distance via contrastive fine-tuning of DNABERT-2, improving similar-sequence search accuracy.

Protein-protein binding affinity prediction from sequence alone, pairing frozen protein language model embeddings with gradient-boosted trees.

Protein deimmunization framework that aligns a protein language model with GRPO to design sequences carrying fewer MHC Class I T-cell epitopes.

Single-cell multimodal LLM generating natural-language descriptions of cell type, tissue, disease, and pathway activity from scRNA-seq profiles.

mRNA language model with a hyperbolic prediction head encoding the codon-amino-acid hierarchy, beating Euclidean baselines on 9 of 10 property tasks.

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

De novo VHH nanobody binder design that conditions AlphaFold-Multimer with structural templates and language model sequence priors, no retraining.

Cross-species RBP-RNA binding site predictor that turns RNA-binding protein conservation into label smoothing, reaching 0.85 AUC from human to mouse.

Protein solubility mutation-effect predictor built on an anti-symmetric Siamese geometric graph network trained on deep mutational scanning data.

TCR-antigen binding prediction that adapts ESM-2 with LoRA on antigen-specific receptors and fuses the embeddings with a bipartite interaction graph.

Reference-guided anatomical segmentation for medical images, pairing vision-language spatial reasoning with a co-trained SAM 2 mask decoder.

Multimodal drug-target interaction model aligning SMILES, molecule text, taxonomy, and protein sequence with a Gramian volume contrastive objective.

Histopathology model predicting homologous recombination deficiency from H&E slides in ovarian cancer, reaching 0.846 AUC and 0.938 specificity.

Structure-free RNA-small molecule binder discovery model that predicts ligands and their binding sites from RNA sequence using an RNA language model.

Spectroscopy-grounded molecular foundation model that reads NMR, IR, and mass spectra as text, elucidating structures and generating 3D conformers.

Peptide aggregation predictor that scores amyloid propensity at single-residue resolution from ESM-2 embeddings, reaching 0.918 AUC on Serrano157.