A major challenge in systems biology is to quantitatively understand and model the dynamic topological and functional properties of cellular networks, such as the spatial-temporally specific and context-dependent rewiring of transcriptional regulatory circuitry and signal transduction pathways that control cell behavior. Current efforts to study biological networks have primarily focused on creating a descriptive analysis of macroscopic properties. Such simple analyses offer limited insights into the remarkably complex functional and structural organization of a biological system, especially in a dynamic context. Furthermore, most existing techniques for reconstructing molecular networks based on high-throughput data ignore the dynamic aspect of the network topology and represent it as an invariant graph. To our knowledge the network itself is rarely considered as an object that is changing and evolving. In this proposal, we aim to develop principled machine learning algorithms that reverse engineer the temporally and spatially varying interactions between biological molecules from longitudinal or spatial experimental data. Our approaches will take into account biological prior information such as transcriptional factor binding targets, gene knockout experiments, gene ontology, and PPI. Contrary to traditional co-expression studies, our methods unfold the rewiring networks underlying the entire span of the biological process. This will make it possible to discover and trace transient molecular interactions, modules, and pathways during the progression of the process. We will also develop a Bayesian formalism to model and infer the "dynamic network tomography" - the meta-states that determine each molecule's function and relationship to other molecules, thereby driving the evolution of the network topology, possibly in response to internal perturbations or environmental changes. Using these new tools, we will carry out a case study on time series gene expression data from organotypic models of breast cancer progression/reversal to gain insight into the mechanisms that drive the temporal rewiring of gene networks during this process. Finally we will also deliver a software platform offering the tools developed in this project to the public. So far, there has not been work done to consider temporally and spatially varying biological interactions under a unified framework. Our proposed work represents an initial foray into this important problem. Our proposed work represents a significant step forward over the current methodology. We envisage a new paradigm that facilitates: 1) Statistical inference and learning of gene networks that are evolving over space and time, possibly in response to various stimuli and possibly mediating genome-environmental interactions. 2) Thorough exploration of the underlying functional underpinnings that drive the network rewiring, dynamic trajectory, and trend of functional evolution. 3) Uncovering transient events taking place in the dynamic systems, building predictive understanding of the mechanisms of gene regulation, network formation, and evolution. 4) Fast and accurate computational algorithms, with stronger statistical guarantee and greater scalability and robustness in large-scale dynamic network analysis. 5) A full spectrum of convenient software packages and user interfaces for dynamic network analysis, available to the public.

Public Health Relevance

We propose a systematic attempt on methodological development for the largely unexplored but practically important problem of time- and space-varying (rather than static) gene network learning, and Bayesian inference of the latent dynamic multi-functionality of genes/proteins in the context of such evolving networks. We intend to apply our method to the study of the dynamic genome-microenvironmental interactions during breast cancer progression and reversal. Since any complex biological processes such as development and disease pathogenesis involve intricate and transient regulatory events and signal transduction, it is unreasonable to assume that the underlying network of gene interaction is invariant throughout the process. But modern experimental and computational methodology is not able to identify such time/space specific network due to technical difficulty, therefore our proposed new methods for inferring evolving network, and the functional underpinnings behind network rewiring are not only needed, but also necessary;but it is beyond the grasp of convention methods and requires the methodological innovations we propose. Unraveling and characterizing such dynamic activities and trajectories of biological networks can provide a more comprehensive genetic and molecular view of complex biological processes and diseases, which may lead to better understanding of the mechanisms driving genome-microenvironmental interactions, and identification of key elements in the network responsible for the functional integrity of the network and the system;in addition, such an approach will allow us to formulate hypotheses regarding the roles of these genes with respect to cell differentiation and disease pathogenesis, and to develop improved diagnostic biomarkers and treatment scheme. PHS 398/2590 (Rev. 09/04, Reissued 4/2006) Page Continuation Format Page

National Institute of Health (NIH)
National Institute of General Medical Sciences (NIGMS)
Research Project (R01)
Project #
Application #
Study Section
Biodata Management and Analysis Study Section (BDMA)
Program Officer
Brazhnik, Paul
Project Start
Project End
Budget Start
Budget End
Support Year
Fiscal Year
Total Cost
Indirect Cost
Carnegie-Mellon University
Schools of Arts and Sciences
United States
Zip Code
Marchetti-Bowick, Micol; Yin, Junming; Howrylak, Judie A et al. (2016) A time-varying group sparse additive model for genome-wide association studies of dynamic complex traits. Bioinformatics 32:2903-10
Lee, Seunghak; Kong, Soonho; Xing, Eric P (2016) A network-driven approach for genome-wide association mapping. Bioinformatics 32:i164-i173
Wang, Xuefeng; Xing, Eric P; Schaid, Daniel J (2015) Kernel methods for large-scale genomic data analysis. Brief Bioinform 16:183-92
Kolar, Mladen; Liu, Han; Xing, Eric P (2014) Graph Estimation From Multi-Attribute Data. J Mach Learn Res 15:1713-1750
Xing, Eric P; Curtis, Ross E; Schoenherr, Georg et al. (2014) GWAS in a box: statistical and visual analytics of structured associations via GenAMap. PLoS One 9:e97524
Parikh, Ankur P; Wu, Wei; Xing, Eric P (2014) Robust reverse engineering of dynamic gene networks under sample size heterogeneity. Pac Symp Biocomput :265-76
Parikh, Ankur P; Curtis, Ross E; Kuhn, Irene et al. (2014) Network analysis of breast cancer progression and reversal using a tree-evolving network algorithm. PLoS Comput Biol 10:e1003713
Yin, Junming; Ho, Qirong; Xing, Eric P (2013) A Scalable Approach to Probabilistic Latent Space Inference of Large-Scale Networks. Adv Neural Inf Process Syst 2013:422-430
Puniyani, Kriti; Xing, Eric P (2013) NP-MuScL: unsupervised global prediction of interaction networks from multiple data sources. J Comput Biol 20:892-904
Curtis, Ross E; Kim, Seyoung; Woolford Jr, John L et al. (2013) Structured association analysis leads to insight into Saccharomyces cerevisiae gene regulation by finding multiple contributing eQTL hotspots associated with functional gene modules. BMC Genomics 14:196

Showing the most recent 10 out of 20 publications