CHIPPENHAM, U.K.—Analytical solutions provider Applied InSilico recently announced that GlaxoSmithKline (GSK) signed a multi-year licensing agreement for the application of its Evolutionary Learning Environment (ELE). The deal is part of a larger GSK initiative to develop a new global decision support environment for drug discovery areas such as model building, library design, drug candidate selection. No financial details were disclosed.
According to Jay Perrett, Applied InSilico CTO, the ELE improves the efficiency of resident computing systems by providing computational resources as part of a background parallel computing network. The system provides data to a "problem server", which distributes the query or calculation to clients along with eDNA (electronic DNA) that is used to "code" a particular algorithm. In a way, the computer network then "learns" to solve the complex data problem.
From GSK's perspective, the new system should allow the company to run more data analysis scenarios in parallel, provide insight on key variables and data relationships, simplify complex models and provide a model environment that continuously learns from experience and data updates. Combined, these parameters should improve drug discovery.
The deal is just the latest for GSK, which has made significant investments in its informatics infrastructure over the last couple of years, including a deal with InforSense and SpotFire in May and several database licensing deals with companies like Prolexys, GeneGo, and Jubilant Biosys. Officials at GSK could not be reached for comment, but in November 2004, GSK Vice President of Cheminformatics Stephen Calvert put the company's interests in perspective.
"Our scientists have to deal with growing quantities of data," he said. "The more efficient we are at processing these data, the better will be our decision making. The challenge is both to facilitate this dynamic and iterative process and at the same time capture the key decisions that are made so they can be improved over time. An organization like GSK must also ensure that the IT infrastructure we build is both scalable and maintainable."
Deals such as this one are becoming more prevalent of late as the pharmaceutical industry is faced with a wealth of data from initiatives like the Microbial Genome Project and the HapMap Project but a dearth of viable bioinformatics and computing solutions. According to Perrett, this dilemma arose to some extent from the victory of hype over substance.
"The fact that a solution could be 'visualized' did not mean the solution could be realized," he explains. "For example, it's fairly easy to describe complex data analysis and automated systems, but the devil really is in the details and bio- and cheminformatics companies seemed to have rehashed old solutions, which was not what customers actually needed."