Yuanchun Zhou

Cell-Graph Compass (CGC) is a foundation model that uses graph structures to represent single cells, capturing multiple biological perspectives and achieving state-of-the-art results in downstream tasks.

scCompass provides a cross-species scRNA-seq database with standardized preprocessing, enabling AI-ready data for research and analysis.

This study evaluates seven large language models on their ability to understand genes and cells, finding they have some capability but require improvements in cell annotation tasks.

GeneCompass is a cross-species foundation model pre-trained on single-cell transcriptomes from human and mouse, integrating biological knowledge to improve gene regulatory mechanism understanding and downstream tasks.

CellPolaris uses transfer learning to generate cell-type-specific gene regulatory networks from transcriptomic data, achieving state-of-the-art results in predicting regulatory factors and simulating cell fate transitions.