Open Projects
All projects
First-author / co-first open projects — filter by category and year.
VenusFactory2
Free web platform with AI agents for protein discovery and directed evolution — integrating Venus / Prot* models in the browser.
VenusRAR
Two-stage agentic framework that ensembles PLMs then audits candidates with biophysical chain-of-thought reasoning.
VenusX
A large-scale benchmark for residue-, fragment-, and domain-level protein functional annotation and pairing.
TCM AI
Open catalog of TCM LLM papers, models, benchmarks, datasets, and industry news — synced with Awesome-TCM-LLM.
VenusFactory
Data retrieval and language-model fine-tuning in one open toolkit for protein engineering workflows.
VenusREM
From high-throughput evaluation to wet-lab studies — advancing fitness prediction with MSA-aware retrieval.
ProtSSN
Semantical and geometrical protein encoding toward enhanced bioactivity and thermostability — no MSA required.
VenusVaccine
Dual-attention multimodal antigen encoding to classify protective vs non-protective antigens.
ProtSolM
Fusion of sequence, structure, and physicochemical features for solubility prediction.
ProtLOCA
Homology detection and local structure matching without amino-acid sequence embeddings.
ProSST
A pretrained protein sequence–structure Transformer with structure tokenization and disentangled attention.
SES-Adapter
Simple, efficient, scalable structure serialization + cross-modal attention for stronger PLM fine-tuning on fitness tasks.
PETA
Evaluating how sub-word tokenization shapes protein pretraining and downstream transfer.
MedChatZH
Baichuan-7B tuned on TCM books and medical instruction dialogues for Chinese clinical Q&A.