Many biological processes are the result of complex events including cascades of transcriptional programs and numerous molecular interactions. Traditionally, research focus has been given to the characterization of individual mutants, regulators or interactions. However, many phenotypes and behaviors cannot be attributed to isolated components. Rather, they arise from characteristics of biological networks, which represent connections between molecules. With the availability of complete genome sequences and high-throughput experimental techniques, probing biological questions on a system level has become feasible. Pioneering work in recent years has provided first drafts of catalogs of essential components, transcriptional regulatory diagrams and molecular interaction networks underlying certain biological processes.
We use C. elegans embryonic development as a model to obtain system-level pictures for biological processes. Our goal is to identify all the involved components, elucidate the organization of these components, and provide local models for protein/gene functions. To achieve this goal we collect datasets emerging from various genome-scale approaches such as DNA microarray, protein-protein interaction mapping, systematic RNAi analysis and gene expression localization mapping. Since each of these genome-wide strategies approaches biological questions from a different perspective and each has its own caveats, datasets generated by different strategies need to be integrated in the context of development. We are developing novel computational approaches to integrate various data sources, construct network models, and identify functional modules from networks, which will eventually yield biological insight to development. We will experimentally validate the functional predictions we make based on genomic data integration.
Right now we focus on the following projects:

 A.  Pleiotropy

Pleiotropy refers to the phenomenon of a single gene controlling multiple distinct and seemingly unrelated phenotypic effects. We investigate the occurrence and possible mechanism of this phenomenon in the context of C. elegans early embryonic development. Using pre-defined functional categories as seeds, we identify the sets of phenotypic descriptors, or “signatures” of defects, which best represent each functional category. We identify pleiotropic proteins with these signatures. Since many cellular events in early development may be mediated by protein-protein interactions, we examine the properties of pleiotropic proteins in interactome networks. We propose that these proteins act as “information exchange centers” between different protein complexes and pathways, which is a fundamental reason for their loss-of-function phenotype complexity.

B.  Phenotypic Robustness

Phenotypic robutstness is achieved in face of genetic variability and environmental changes. One major difficulty in studying development is that many single perturbations do not give rise to any defective phenotypes. One of the major reasons underlying this robustness is that functionally redundant or overlapping genes buffer one another during development. Some functionally redundant or overlapping genes are similar in sequence whereas others are not. By investigating the topological features of interactome networks and integrating other types of high-throughput data, we predict potential gene pairs/groups in the network that buffer the functions of one another during C. elegans embryonic development. We use dual RNA interference analysis to verify these predictions.

C.   Network Dynamics

Biological networks are dynamic rather than static. However, due to technological limitations, currently many descriptions of biological networks are compositions of interactions that may not occur simultaneouly. We are interested in obtaining a dynamic description of networks by integrating molecular interactions with expression datasets which provide information for location/condition/time series. We identify network structures that are temporally/spatially important for development.