Global analysis of biological function modeling and prediction

Understanding cellular processes is very important in post genomic era. Systems biology studies organisms, and their interacting networks of genes, proteins and biochemical reactions. My research focuses on formulating optimization problems and proposing algorithms for studying global cellular processes and molecular functions in context of large scale biological networks including genetic regulatory networks, transcriptional regulatory networks and protein-protein interaction networks.;In predicting protein function study, two combinatorial algorithms based on connected graphs of PPI network are proposed. The developed reliable path searching algorithm can be applied to find reliable proteins for function annotation, and can also be further applied to extract disease related protein from the physical protein interactions.;In the study of inferring functional linkages among proteins for function annotation, a global analysis method is presented to identify the biological modular structures via the consensus pattern extraction from protein set and species set simultaneously. It aims to reveal the functional linkages with high confidence in the extracted modules.;In gene regulatory network study, two-level-simulated-annealing method is formulated and can optimally solve the problem of searching the optimal Bayesian network structure with a minimum score. The method outperforms other heuristic searching methods in terms of the optimal score.;Network motifs are building blocks in many biological networks. The study focuses on finding statistically significant motifs with edge-joint property. A parallel method is developed to detect large size network motifs, some of which have been verified via "wet-lab" experiments in the yeast and E. coli transcriptional regulatory networks.;All the above four problems are centered around the key issue of revealing biomolecular functions, which is one of the major goals of systems biology. The work tackles the problem from different level of biological networks and perform global analysis of bio-molecule interactions based on multiple high throughput biological data. The contributions are mainly on the improvement of the computational methods on above biological methods. These methods can be used as efficient tools to the progress of understanding the potential drug targets or the new design principles of biological networks in the fields of systems biology and bioinformatics.