These themes define how we present the lab externally. Collaborative multi-omics studies related to neuroscience are documented on the Publications page only and are not featured as a main research direction.
Chromatin folds into hierarchical 3D structures that shape gene expression. We develop deep-learning models that transfer across cell types and experimental settings to predict interactions among accessible regions, multi-scale chromatin loops, and TAD boundaries. Reference implementations for published methods are released on GitHub.
Enhancers and open chromatin form complex cis-regulatory networks. The lab builds interpretable deep networks and transfer-learning frameworks for open-chromatin prediction, super-enhancer versus typical-enhancer classification, and enhancer-enhancer interaction modeling.
Multi-omics analysis of complex disease to identify key regulatory factors and biomarkers, together with deep-learning applications in medical diagnosis, biomedical imaging, and precision medicine.
Model objectives are grounded in real biological questions, with emphasis on verifiable performance and interpretability.
We prioritize transfer learning and multi-scale modeling so methods remain robust when data conditions change.
Code for published methods is released on GitHub so collaborators can reproduce and extend the tools.
We welcome students and collaborators with backgrounds in computational biology, bioinformatics, computer science, biomedical engineering, and related fields.
Contact the lab Browse papers