The research in our group is focused on studying the biological principles using statistical and biophysical modeling. We are particularly interested in reconstructing the intracellular regulatory networks, understanding the biophysical and biochemical mechanisms through which the biological regulations are achieved, and uncovering the principles that govern the evolution of these networks.

A simplified sketch of such regulatory networks is shown in the figure on the right. Diverse environmental changes are detected, causing signals to be transduced through signaling pathways. Particular transcription factors are then activated and transported into nucleus to transcribe their target genes. Protein products of these genes interact with other proteins in the same or other signaling pathways to further tune responses to extracellular stimuli, thus producing a variety of feedback loops.

 
The current ongoing projects in the lab include:
 
1. Reconstruct the transcriptional network. To determine the links in the transcriptional network, one needs to identify the regulatory element recognized by each transcription factor in the yeast genomeand the target genes regulated by each transcription factor as well as to understand the cooperation between transcription factors (combinatorial regulation) in a condition-dependent manner. We have developed and are continuing improving computational methods to analyze the genomic data such as gene expression experiments and ChIP-chip experiments to identify binding motifs and target genes of transcription factors. We currently are mainly working on the budding yeast, on which there are many high-throughput data available. These methods will be extended to human in the future.
   
2. Determine the binding specificity and protein interacting partners of peptide-recognition protein domains. Proteins has a modular design in the sense that they often contain independently-folded domains that perform specific functions. Peptide recognition domains, such as SH3, SH2 and PDZ domains, can binding to specific amino acid sequences and often play significant roles in signal transduction. The binding between the domains and the peptides are usually weak and transient and thus are difficult to be detected by high throughput technologies. We have been developing computational methods that combine information from biophysical study such as molecular dynamics and free energy calculation and bioinforamtics analysis such as sequence alignment to predict the binding sequence and thus the interacting partners of these modular domains as well as to understand the biophysical mechanisms that determine the binding specificity.