Xujing Wang, PhD
Dept. of Physics
School of Natural Sciences and Mathematics
Office Address: Campbell 310
Websites: Wang Lab
NanKai University, China
BS, Physics, 1988
Texas A&M University, College Station, TX
PhD, Theoretical Physics, 1995
University of Texas MD Anderson Cancer Center, Houston, TX
Postdoctoral Fellow, Biophysics and Medical Engineering, 1995-9
The genetic research of my groups has mainly focused on the following areas:
1. Network Biology
Presently, my group focuses on the modeling of transcription regulation networks from gene expression data. The interest include the identification of network module/motifs; study of network structure versus function, and the dynamics under pathological/physiological changes; integration with clinical and phenotypic data for mechanistic investigation of disease pathogenesis, and development of disease markers (expression signatures). More specifically we have been investigating the dynamic Bayesian Network (BYN) approach. We introduced fuzzy theory based rules in the simulation, to avoid local minima; designed and implemented algorithms to incorporate prior knowledge including protein-protein interaction, ontological similarity and PubMed cocitation, to assist the determination of initial and candidate network structures. These lead to improved performance over naïve BYN approaches. Additionally, we are developing algorithms specific for time course data, utilizing techniques including the phase synchronization analysis and the Granger causality. These approaches can better address the timing lag issues in transcription regulation. We plan to continue these developments.
With existing collaborators we are investigating expression signatures for the initiation of adaptive versus innate immune responses during type 1 diabetes (T1D) and asthma. Most of our developments, though initially made for transcription networks, can be extended to other types of networks including the protein-protein interaction and metabolic networks. Further, network modeling is not necessarily limited to one type of data, integration of data from different biological scales will allow a systems study of network biology. We are specifically interested in the integration of phenotypic and genetic/genomic data, to discriminate the causal genetic variation, primary phenotypic changes, from secondary genetic (expression) or phenotypic changes.
2. Integrative Genomics of Complex Disease
Complex human diseases typically result from the interplay of multiple, interacting genetic factors. Therefore understanding the disease biology is much needed to dissect the genetics risk. My group has been developing a multi-level, integrative genomics approach, and applying it to diabetes. It first investigates and identifies key quantitative traits and disease pathways that are important to the disease initiation, through the dynamic modeling of disease pathogenesis. It then studies the network structure of genes involved in these pathways. Based on the results, it compiles a comprehensive list of candidate genes, and uses a Bayesian classifier to prioritize the candidate genes. When applied to T1D, it led to the identification of many known disease genes, as well as prediction of new candidates. Presently with collaborators we are typing the new predictions in a cohort that we have obtained from Finland, and a replicate cohort from the Type 1 Diabetes Genetics Consortium (T1DGC). This project is currently funded by NIDDK/NIH through Oct of 2011 (R01 DK080100-01).
In our future plan we will also incorporate the recent GWAS (Genome Wide Association Study) data in the identification of disease pathway and candidate gene prioritization. In the mean time, our approach can be directly applied to GWAS analysis. At the moment, only markers with extremely low p-value (usually <~10-7) are retained. Lowering the threshold will be plagued with false positives, though it is believed that a region immediate below the threshold p value likely also harbors many true disease genes. We plan to develop analysis algorithms to discriminate between true disease genes from false positives in this region, and to identify the etiological variants among markers in LD. Our integrative genomics approach, by design, is applicable to the other diseases. With collaborators we plan to look into type 2 diabetes, cardiovascular diseases, and asthma.
Other interest: Systems Biology of Glycemic Control
Glucose homeostasis is a fundamental physiological process that provides energy to all cells in the body. To maintain the blood glucose concentration within the narrow physiological ranges it takes multiple hormones and several tissue organs to operate synchronously at multiple levels. we are developing a multi-scale (include intracellular, cellular, interceullar, islet/pancreas, and blood circulation), systems approach that incorporates both spatial (tissue structural organization, etc) and temporal (insulin dynamic rhythms, etc) considerations, to investigate insulin secretion regulation, its role in glycemic control, and changes responding to pathological modifications such as diabetes. One particular question we are focusing now within this framework is the nonlinear relationship between β-cell function and β-cell mass. We have, for the first time, characterized quantitatively the functional role of the architectural organization of islet β-cell mass. We are now collaborating th islet transplantation investigators, and plan to develop predictive models for islet function and survival, and QC for transplantation. In addition, we also plan to develop predictive models of β-cell mass from functional (insulin secretion) measurements, this will allow early detection of the β-cell destruction during diabetes. This includes the extension of MINMOD in collaboration with Dr. Richard Bergman at USC, and integration with our studies at lower biological scales. Together with experimental collaborators we are planning rodent studies to test our model predictions. In the future we plan to translate such work clinically to develop early disease markers for T1D, and to assess the (residual) β-cell mass at diabetes onset. On the technical side, we are working with collaborators to develop better numerical methods to facilitate these modeling efforts.
DRC Membership Category: