Single-cell Omics
Method Development
Single-cell sample-level embedding
- representation learning
- scRNA-seq
- method development
- statistical simulation
- random matrix theory
Overview
Developed a novel sample-wise similarity metric to enable high-level embedding of scRNA-seq data, allowing researchers to compare entire patient cohorts rather than just individual cells.
Key highlights
- Designed statistical simulations to ensure metric robustness against batch-effect perturbations.
- Applied the method to infer population structures and identify specific gene distribution shifts driving biological transitions.
- Facilitated cross-sample comparisons to discover latent trajectories in large-scale transcriptomic datasets.
