The long term goal of the CompBio research focus area is reconstruction of gene regulatory networks. We are developing new algorithms, tools, and databases that enable us to identify and characterize the components of the networks, to discover interactions among the components, and to infer the structure of the networks. In addition, we are developing new capabilities in the areas of data provenance, visualization, and co-processor acceleration to support the modeling process. Our research integrates algorithms and methods for:
- Identification and characterization of genes/proteins. Massively parallel sequencing and and high throughput proteomics are dramatically increasing the data available for building biological networks, but the computational tools for processing and mining this data are immature. We are developing new algorithms and pipelines for extracting and integrating information about the genes that are expressed under specific conditions, the effects of DNA modifications and microRNA on the expression of genes and translation to protein.
- Determining interactions of the components. Although substantial information about biological pathways is available for model organisms such as yeast, mouse, human, this information is largely lacking for non-model species. We are developing new approaches to extract and predict the reference pathways for organisms where pathways are not available.
- Network reconstruction. The challenge facing researchers in building gene regulatory networks from data is to develop methods for reconstructing the networks that exist in nature. We are investigating improved approaches for inferring networks from data based on information theory, equivalence classes of models, probabilistic Boolean models and dynamic Bayesian models.