Beijing Zhongguancun Academy

Multi-label therapeutic peptide classifier over frozen ProtT5 embeddings, scoring 15 bioactivities from antimicrobial to quorum-sensing in one pass.

EEG foundation model coupling spatial and temporal transformer branches through a shared soft mixture-of-experts, adapted by tuning 5.1% of weights.

All-atom foundation model for immune-receptor design that predicts structures and co-designs CDR sequences for antibodies, nanobodies, and TCRs.

860M-parameter generative single-cell foundation model that jointly represents and generates epigenomic, transcriptomic, and proteomic modalities.

Autoregressive DNA foundation model for variant effect prediction, using 6-mer tokenization to match Evo2-7B win rates at far higher throughput.

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

Structure-based drug design model that generates ligands for a protein pocket, pairing a diffusion structure encoder with preference optimization.

Protein foundation model with 3B parameters, pretrained jointly on sequence and 3D structure via masked language modeling and diffusion denoising.

Hierarchical transformer for virus discovery in metagenomes, classifying viral genomes across taxonomic ranks and flagging candidate novel lineages.

Compact 1.1M-parameter DNA language model distilled from Nucleotide Transformer v2, outperforming its 500M teacher on 11 of 18 benchmark tasks.

Long-context generative genomic foundation model with a 98k-nucleotide window, trained on 386 billion bases of eukaryotic DNA for sequence design.

Prokaryotic promoter design and zero-shot promoter-strength ranking from a nucleotide language model pretrained on 17,806 microbial genomes.