Google Algorithm Sorts Chemical Links For Scientists
Google's search method turns out to be useful for more than just finding relevant Web pages. It works just as well on chemical links, which can help scientists design new medicines for treating disease.
While Google relies on PageRank to help it evaluate which Web pages are relevant for a given search query, Aurora Clark, an associate professor of chemistry at WSU, has been using the algorithm to predict chemical reactions.
White PapersMore >>
PageRank measures the relative importance of a set of Web pages connected by hyperlinks. Web pages with large numbers of incoming links are deemed by the algorithm to be more important than those with fewer incoming links. In effect, the act of creating a link to a website becomes a vote for the importance of that site.
"The way in which Google tallies votes of confidence for Web pages is based on their connectivity," said Clark in a phone interview. "The connectivity [as measured by the number of links to the Web page] dictates the probability that the Web surfer will stay at, or leave, the Web page. We're doing the same thing."
[ Google's DNS business is providing the company with even more insight into what people do online. Read Google Becomes Largest Public DNS Provider. ]
Clark says that the number of hydrogen bonds in a water molecule makes it more likely to be relevant for chemical reactions.
"What the PageRank algorithm provides is a way to characterize the organization of a system all at once," she said.
The research, funded by the U.S. Department of Energy, is described in a paper published online in The Journal of Computational Chemistry, "moleculaRnetworks: An Integrated Graph Theoretic and Data Mining Tool to Explore Solvent Organization in Molecular Simulation," co-authored by Clark, Barbara Logan Mooney, and L. Rene Corrales.
moleculaRnetworks refers to a set of scripts written in the R computer language for identifying, visualizing, and understanding molecular reactions. PageRank is one of the algorithms used in the scripts.
Understanding how molecules are organized can help with the design of medicines and with the treatment of diseases. "There's a large branch of chemistry about how molecules freeze, melt, and transform, all based on molecular organization," Clark said.
Clark's work focuses on toxic and heavy metals, such as uranium and plutonium, and how they behave in water. Having a computational tool to explore the reactivity of these elements in water expands research options without the challenge of dealing with these dangerous substances directly.
Google, understandably, is pleased that its founders' seminal work has cross-disciplinary applications. "Our goal in search is to help people expand their knowledge of the world, and we're delighted to see that our PageRank algorithm is being used to do just that with this innovative and efficient molecular research method," said Amit Singhal, Google Fellow and senior vice president, in an emailed statement.
The right forensic tools in the right hands are just a start. The new Digital Detectives issue of Dark Reading shows you how to better apply the lessons they teach. (Free registration required.)