Tongji University

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

Sequence-conditioned generative framework that reconstructs missing prostate MRI contrasts and restores artefact-degraded acquisitions.

Renal tumor histopathology model that detects tissue regions, classifies nine subtypes, grades nuclei and scores prognosis from a single H&E slide.

Single-cell foundation model adding a gated cell-level contrastive objective to masked expression pretraining for transferable frozen cell embeddings.

Motion artifact reduction for cardiac CT, folding a learned cardiac motion field into model-based reconstruction to image the heart at any phase.

Connects a frozen protein language model to a general LLM via a cross-modal projector, adding protein reasoning without catastrophic forgetting.

De novo peptide sequencing transformer that reads modified and unmodified peptides directly from tandem mass spectra without a reference database.

Single-cell foundation model adapting LLaMA-3.1-8B with LoRA, recasting transcriptomes and protein interaction networks as natural-language Q&A pairs.

Gene expression prediction from candidate cis-regulatory elements, fusing DNA sequence, TF binding scores, and epigenomic tracks in a CNN-transformer.

Tissue-specific RNA splicing prediction from pre-mRNA sequence, scoring how variants shift splice-site usage across 18 human tissues.

Multimodal drug-response model coupling cell and molecule foundation models, pretrained on 1.8M perturbation RNA-seq profiles over 22,000 compounds.

RNA-protein contact prediction from sequence, built on ERNIE-RNA and ESM-2 embeddings. Reaches 0.77 auROC where AlphaFold 3 reaches 0.61.

Relative protein-ligand binding affinity prediction from docked complexes, matching Schrodinger FEP+ ranking accuracy zero-shot on the FEP benchmark.

EEG foundation model that separates channel-wise from temporal attention, pretrained on 25,000 hours of recordings spanning eight task paradigms.

Regulatory genomics foundation model pretrained on 6,391 human ChIP-seq cistromes, representing how ~1,000 transcription regulators cooperate.

Secondary metabolite structure prediction from microbial biosynthetic gene clusters, generating SMILES strings from Pfam functional-domain tokens.

Antimicrobial peptide discovery from metagenome-assembled genomes, labelling AMP residues with a LoRA-adapted ESM-2 token classifier.

Protein conformational ensemble generator using SE(3) flow matching from a perturbed ESMFold prior, sampling MD-like dynamics from sequence alone.