Fully Loaded

COVID-19 Community Update

For the latest updates & information visit Duke’s Coronavirus Response website.

News & Events

Now in Print

Now in Print is a series highlighting recent publications by StatGen’s members. If you want to learn more about a particular article or researcher, please contact us at statgen@duke.edu.

Recently, we sat down with Jichun Xie, PhD  to learn more about one of her current research projects and two associated publications.

What are your current research interests?

[JX]:  My research focuses on identifying rare events and inferring complex dependence structures of high dimensional data, and their application in translational biomedical research. This work explores the identification of complex gene co-expression networks based on gene expression data and their special structures. Current projects aim to identify rare variants in heterogeneous genetic and genomic data with a focus on gene regulatory networks based on gene expression data and rare variants in heterogeneous high-throughput sequencing data.


What recent publications highlight this area of research?

[JX]:  Two recent publications establish new methodological approaches in this area.  In “PenPC: A two-step approach to estimate the skeletons of high-dimensional directed acyclic graphs”, we infer the skeleton of a high-dimensional directed acyclic graph (DAG) of gene expression.  DAG is commonly applied to model causal relationships between gene expressions and the PenPC method provides a computationally efficient method in estimating its skeleton with effective asymptotic properties of convergence.  The method also offers higher sensitivity and specificity than the previously established, state-of-the-art method, the PC-stable algorithm.

A second recent publication, “Joint estimation of multiple high-dimensional precision matrices” , jointly infers the Gaussian graphical models of gene expressions in multiple tissues/subpopulations, by assuming the graphical modes are similar but not identical across tissues or subpopulations.  This joint estimation method improves the accuracy in network inference compared with the separate estimation method and the pooled estimation method. Theoretically, the more tissues/subpopulations included in the data, the greater the improvement. Simulation results also shows that it outperforms other competitive joint estimation methods.

What are the next steps to further explore this topic?

[JX]:  The next steps include exploring methods for gene expression data, such as measured by RNA-seq, which cannot always be normalized to Gaussian distributions.  The association pattern between genes could also be complicated, offering additional challenges.  To address these issues, we are developing a testing method to infer high-dimensional general association networks of gene expressions conditioning on covariates.


Dr. Xie joined the Department of Biostatistics & Bioinformatics as an Assistant Professor in July 2014 and has been involved with StatGen since its inception.  Her research extends beyond StatGen with an affiliation with the Duke Cancer Institute and collaborations with the Department of Neurosurgery.  Dr. Xie also teaches several PhD and MB courses in the Biostatistics & Bioinformatics department and serves as a mentor and committee member for students in both programs.

To learn more about Dr. Xie’s research or how to collaborate, please contact StatGen at statgen@duke.edu


Ha, MJ, Sun, W, and Xie, J. “PenPC: A two-step approach to estimate the skeletons of high-dimensional directed acyclic graphs.” Biometrics 72, no. 1 (March 2016): 146-155. [Scholars@Duke] [PubMed]

Cai, TT, Li, H, Liu, W, and Xie, J. “Joint estimation of multiple high-dimensional precision matrices.” Statistica Sinica (2016). [Scholars@Duke] [PubMed]