Jialiang Huang

The paper introduces eScreen, a deep learning model that decodes regulatory genome function at single-nucleotide resolution using CRISPR screen data, achieving high accuracy and mechanistic interpretation.

Super-CUT&Tag improves spatial epigenomic profiling by increasing sensitivity and resolution, revealing active-chromatin heterogeneity in mouse embryonic tissues.

This study benchmarks five loss-of-function perturbation tools for studying noncoding genome function, finding dual-SpCas9 is most effective for targeting CREs and recommending RNA interference for noncoding RNA genes.

Chrombus-XMBD predicts 3D-genome structures from chromatin features using a graph generative model, outperforming existing methods and showing cross-species generalizability.

The paper identifies prognostic-associated subpopulations and develops recurrence risk signatures from scRNA-seq data of colorectal cancer using a rank-based framework.

The paper integrates single-cell transcriptome data from liver samples to identify cell lineage similarities and prognostic subpopulations between cirrhosis and liver cancer.

The paper introduces eNet, a method to build enhancer networks from single-cell data, showing complex networks regulate cell identity and disease genes better than existing models.

The paper develops a new method to map nuclear pore-chromatin interactions and finds NPCs are associated with super-enhancers and form transcriptional condensates.

The paper shows that enhancer-dependence of gene expression increases with developmental age in mouse and human cells.

The paper identifies hub enhancers within super-enhancers using Hi-C and ChIP-seq data, showing their role in gene activation and chromatin structure.