 |
 |
|||||||
|
||||||||
Kuffner, R.M., Gonzales, M., Steadman, P., Wlodek, D.K., Jankowitz, R.J., Boinoff, J.R., MontoyA, L., Peterson, T.F., Bulmore, D.L., Blanchad, J.B.
National Center for Genome Resources 2935 Rodeo Park Dr. Santa Fe, NM USA 87505
Contact jlb@ncgr.org
What is the difference between two cells? Intuitively, the answer will be dependent on the genotype, development, location, and current environment of the cells. The answer will also be dependent on integrating knowledge derived from many different chemical and biological experimental data sets. To facilitate research we have developed a curated relational database of information on the mapping or interaction data between cellular building blocks (i.e. genes, proteins and metabolites) in order to build and analyze complex cellular models. We also curate data describing how these building blocks are modified and regulated. This interaction data is derived from the literature, sequence annotation and from analytical methods (i.e. prediction of an ATP binding domain or a regulatory site). Our content development efforts are currently focused on Arabidopsis, yeast and mouse. The Arabidopsis data set is comprised mainly of interactions between proteins and metabolites, while the yeast data set also includes annotation derived from other public databases, protein-protein, protein-DNA and protein-lipid data sets. We have only recently begun curating mouse data. The data model includes many features of the Gene Ontology model for easier navigation among biological classes. Using the PathDB QueryTool cellular building blocks can be transformed into a set of interactions (or transformed directly to the set of canonical pathways associated with the building blocks). Next a pathway or network model is constructed from the set of interactions. By using the "Neighborhood" method to retrieve building blocks from PathDB known to interact with building blocks in the original data set the pathway model can be sequentially expanded. In this fashion complex models can be assembled. The PathwayViewer is capable of presenting a single view of a network composed of metabolic, signaling and other types of "classical" pathways. The analysis of complex models built de novo can be quite difficult. For instance what are novel vs. known pathways or routes? One approach is to allow the users to overlay and highlight other information types. Thus the user can ask which of the protein-protein interactions derived from yeast two-hybrid experiments are present in the literature or genome sequence annotation. Using PathDB information about subcellular location or phenotype information like programmed cell death or disease resistance can be displayed in the context of known biological interactions. A pathway model can also be used to visualize gene, protein and metabolite expression data and the view can be synchronized with other software programs using NCGR's integration system ISYS. Since our analytical tools will help generate new interactions among building blocks we allow users the ability to store the information in the database. We are also developing statistical tools for validating network models. There are two mechanisms for accessing the database. For our research work we use and have made publicly available a Java Client that is integrated with ISYS (For information on ISYS and related gene expression programs see www.ncgr.org/isys). To enhance the utility of the database as an information resource we have developed a simplified html version of the research tool. Both interfaces can be found at www.ncgr.org/pathdb.
PathDB is funded through Grant# DAAH04-96-1-0415 from the Army Research Organization to the National Biotechnology Information Facility at New Mexico State University. We wish to acknowledge the following persons for past contributions to the project: Andrew Farmer, Jerry Kerrisk, Pedro Mendes, Mary O’Connell, Dawn Perry, Adam Siepel, Bruno Sobral, John Spalding, Mark Waugh and Stan Wlodek.
