Shenzhen Institutes of Advanced Technology, CAS

Protein perturbation model conditioning substitution-effect prediction on a learned protein-level regime coordinate over 202 million sequences.

Prokaryotic genome language model that reads annotated replicons as ordered gene-product descriptors to predict plasmid hosts and gene essentiality.

Species-conditioned codon language model that jointly reads 5'UTRs, coding sequence, and RNA secondary structure to design native-like genes.

Cervical cytology screening system pretrained on 127,471 whole-slide images from 48 centers, with test-time adaptation for new clinical sites.

Transformer predicting microbial gene expression from an annotated genome alone, using protein language model embeddings of every coding sequence.

Vision-language model for annotation-free pathology localization, marking the finding a text prompt names in X-ray, histology and fundus images.