BioMap

Structure-aware protein language model using structure-guided masking and a causal objective for variant effect prediction and protein discovery.

Hierarchical DNA foundation model that co-trains a dynamic token-merging tokenizer with latent Transformers to match genomic information density.

Transcriptome-guided diffusion model generating Cell Painting images for unseen perturbations, improving MOA retrieval accuracy by 16.9% over IMPA.

DNA foundation model for germline variant pathogenicity, pretrained on 27 mammalian genomes and fine-tuned on ClinVar and HGMD for SNVs and indels.

Long-sequence DNA foundation model with groove-aware convolutions and reverse-complement gating over 100kb contexts. Averages 0.708 MCC on GUE.

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

Multi-omics foundation model that folds DNA, RNA, and protein into one codon-level nucleotide representation following the central dogma.

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

Scaling-law study of protein language models identifying compute-optimal training for causal and masked objectives on 939 million protein sequences.

Single-cell transcriptomics foundation model with 100 million parameters, pretrained on over 50 million human scRNA-seq profiles for cell embeddings.

Unified 100-billion-parameter protein language model combining autoencoding and autoregressive objectives for protein understanding and generation.

Asymmetric encoder-decoder transformer for single-cell RNA-seq that encodes only non-zero genes, cutting FLOPs 10-100x versus standard transformers.

Protein model quality assessment predicting per-residue lDDT for monomer and multimer interface models from graph-coupled ESM embeddings.