Submitted to: Briefings in Bioinformatics
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: September 26, 2005
Publication Date: December 1, 2005
Citation: Keele, J.W., Wray Jr, J.E. 2005. Software agents in molecular computational biology. Briefings in Bioinformatics 6(4):370-379. Interpretive Summary: Software agents are autonomous computer programs that are capable of goal-directed behavior. In some ways they are like computer viruses except they are constructive and benevolent. Groups of communicating software agents could play an important role in helping genome scientists develop markers for future studies. The advent of genome sequencing in cattle and swine increases the complexity of data analysis required to conduct research in livestock genomics. Genome databases are always expanding and there are inconsistencies in terminology used. Agents developed in different places by different scientists can communicate and cooperate with one another. Even when some of the agents and databases are unavailable (e.g., a temporary system shutdown), agents can at least partially answer user queries. A review of existing systems indicates an agent approach to gathering genomics data should prove fruitful.
Technical Abstract: Communities of software agents could play an important role in helping genome scientists design reagents for future research. The advent of genome sequencing in cattle and swine increases the complexity of data analysis required to conduct research in livestock genomics. Databases are always expanding and semantic differences among data are common. Agent platforms have been developed to deal with generic issues such as agent communication, life cycle management, and advertisement of services (white and yellow pages). This frees computational biologists from the drudgery having to re-invent the wheel on these common chores, giving them more time to focus on biology and bioinformatics. Agent platforms that comply with the Foundation for Intelligent Physical Agents (FIPA) standards are able to interoperate. In other words, agents developed on different platforms can communicate and cooperate with one another. Many software agent platforms are peer-to-peer which means that even if some of the agents and data repositories are temporarily unavailable, a subset of the goals of the system can still be met. Past use of software agents in bioinformatics indicates that an agent approach should prove fruitful. Examination of current problems in bioinformatics indicates that existing agent platforms should be adaptable to novel situations.