Tencent AI Lab

Protein structure autoencoder compressing backbone coordinates into a latent space, paired with a latent diffusion model for generative design.

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

Antibody and TCR CDR sequence design by structure retrieval, matching query loops against solved CDR structures rather than generating residues.

Multimodal medical imaging foundation model for zero-shot clinical diagnosis and report generation from chest X-ray and CT in English and Chinese.

Deep graph contrastive learning framework for single-cell proteomics embedding, handling peptide uncertainty, missingness, and batch effects.

GPT-style DNA foundation model trained on over 200 billion base pairs of mammalian genomes for sequence generation, classification, and regression.

Generative transformer that translates single-cell transcriptomes into proteomes, inferring missing protein abundance from RNA expression alone.

Pretrained transformer for cell type annotation of scRNA-seq data. Trained on 1.1M cells; outperforms supervised methods on cross-dataset transfer.

Histopathology tile encoder pairing a CNN stem with a multi-scale Swin Transformer, pretrained on 15.6 million unlabeled H&E patches.

Rigid protein-protein docking model that predicts a complex from two unbound structures in one pass, with no candidate sampling or refinement.

Histopathology patch encoder self-supervised with BYOL on 2.7M unlabeled H&E tiles, using attention that gates feature channels with pooled context.

Molecular graph transformer pretrained on 11 million unlabelled compounds, used as a frozen fingerprint source or fine-tuned for property prediction.

Pretrained 3D-ResNet backbones for volumetric medical image analysis, co-trained across eight CT and MRI segmentation datasets for transfer learning.