Skip to main content
Loading Events

« All Events

  • This event has passed.

Friday – Statistical Genetics Seminar: Zicheng Ji

April 22 @ 3:00 pm - 4:00 pm

Statistical Genetics Seminar_Zhicheng Ji 0422

Statistical Genetics Seminar

Zicheng Ji, Assistant Professor, Department of Biostatistics and Bioinformatics

Hosted By Yun Li, Professor of Genetics and Biostatistics

Statistical Methods for Decoding Gene Regulation in Single Cells

On Zoom

 

Abstract:

Single-cell sequencing is rapidly transforming biomedical research. With the ability to measure omics information in individual cells, it provides unprecedented resolution to study heterogeneous biological and clinical samples, enabling scientists to discover and characterize previously unknown biological signals and processes carried by novel or rare cell subpopulations.

The new data structure and high level of noise in the single-cell genomic data pose significant analytical challenges. In this talk, I will present methods we developed for analyzing single-cell transcriptome and regulome data. First, I will introduce how to use single-cell RNA-seq data to infer cells’ underlying developmental trajectories using “pseudotime” analysis. We developed TSCAN, a cluster-based minimum spanning tree approach to facilitate the accurate construction of pseudotemporal trajectories by regularizing the complexity of spanning trees. By improving the bias-variance tradeoff of the spanning tree estimation, our method substantially improved the accuracy and robustness of the pseudotime analysis. Second, I will introduce RAISIN, a general regression framework that enables the population-level inference of single-cell RNA-seq data with multiple samples. To model the hierarchical structure of the data and handle small sample sizes or small cell numbers, RAISIN extends linear mixed models with Bayesian shrinkage estimation to regularize variances. Third, I will introduce SCATE, a method to extract and enhance signals from the highly noisy and sparse single-cell ATAC-seq data. This approach can accurately infer genome-wide activities of individual cis-regulatory elements by adaptively integrating information from similar cis-regulatory elements, similar cells, and massive amounts of publicly available regulome data.

 

**Zoom link will be provided through UNC Department of Genetics emails. Contact jcornett@email.unc.edu for meeting link if you have not received. Links will be provided to UNC affiliated email addresses only.

Details

Date:
April 22
Time:
3:00 pm - 4:00 pm
Event Category:
Comments are closed.