The goal of this project is to create bioinformatics software that will rapidly and accurately search and mine proteomics-related information from multiple databases and relate the expression of proteins to upstream or downstream events. The proposed product will link upstream and downstream data to those from multiple biotechnologies (e.g., two-dimensional gel electrophoresis, liquid chromatography, microarray), as well as other files such as ADMET (adsorption, distribution, metabolism, excretion, and toxicology) resources, and present results in a form that will help researchers to understand biologically-significant molecular responses. We propose a web-based proteomics discovery tool using intelligent agents with enhanced search capability based on a semantic similarity measure between ontology terms which takes into account the semantic, syntactic, and structural information available in protein databases. Using the similarity measure, weighted links are induced across ontologies from the textual description, biomedical references, and protein sequences associated with those in protein databases. This approach will make it possible to capture similarities that arise from orthogonal, but de facto closely related, ontology annotations, while resolving the problems of synonymy and polysemy. Intelligent agent technology will be used to continuously search databases for correlations and alert researchers when new relevant information is obtained
