Tsinghua University

A tri-modal foundation model unifying histology images, spatial transcriptomics, and biological language for zero-shot spatial biology and pathology reasoning.

A protein-text foundation model embedding sequences and natural language in a shared token space, enabling protein understanding and de novo design from one checkpoint.

A physics-informed generative foundation model for quantitative diffusion MRI that maps brain microstructure (tensor, kurtosis, NODDI) and adapts zero-shot to each participant's data.

A single-step generative model for protein structure prediction and binder design that reaches AlphaFold3-level accuracy with a claimed ~15x inference speedup.

A co-generative protein language model that jointly decodes sequence and structure tokens from GO functional annotations for de novo functional protein design.

All-atom generative foundation model trained on 5M+ biomolecular complexes that designs small molecules, peptides, and nanobodies against a target site from one checkpoint.

A DeepSeek-7B-based multi-task large reasoning model that applies chain-of-thought reasoning and reinforcement learning across ~10 molecular science task families.

Hierarchical language model for atlas-level cell-type annotation of scATAC-seq data that annotates new query datasets without retraining.

A hierarchical sequence-based protein representation that encodes proteins as discrete 'words' for zero-shot functional discovery and generative design.

Latent diffusion model that designs D-peptide (mirror-image) binders against native L-protein targets via cross-chirality generalization, with wet-lab validation.

A brain foundation model that unifies EEG and MEG signals via a shared discrete tokenizer and self-supervised pretraining on 2,653 hours of recordings.

A hypergraph foundation model for brain disease diagnosis, self-supervised on high-order fMRI connectivity and fine-tuned few-shot across four diseases.

A multimodal LLM that aligns a specialized ECG signal encoder with BioMedGPT-LM-7B for cardiovascular disease detection and ECG question answering.

A multimodal vision-text foundation model for brain CT and MRI, pretrained on ~10M paired images and reports to act as a clinical copilot across seven imaging tasks.

End-to-end framework predicting protein structure and mutational fitness from a single sequence, with 5x faster inference than ESMFold at comparable accuracy.

A structure-enhanced RNA language model that incorporates base-pairing constraints into self-attention, achieving state-of-the-art RNA structure and function prediction.

A 368M-parameter generative language model for single-cell transcriptomics, enabling zero-shot cell type annotation, batch integration, and conditional cell generation.

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

Generative diffusion model for single-cell RNA-seq data synthesis, enabling controlled generation of specific cell types, rare cells, and developmental trajectories.

Transformer model predicting context-specific epigenomic signals across cell types using DNA sequence and transcription factor activity profiles.