Original URL: http://www.theregister.co.uk/2011/12/06/ms_research_hiv/
Microsoft researchers build spam filter for HIV
Turns out spammers behave a lot like deadly virus...
Researchers at Microsoft have discovered that tools first developed to fight email spam can be applied in helping to understand how the process by which HIV mutates to avoid attack by the immune system.
Microsoft Researchers David Heckerman and Jonathan Carlson were asked to help AIDS researchers in Africa to make sense of data from HIV vaccine testing programmes. The data was compiled by a consortium of hospitals and universities, including MIT, the Center for the AIDS Programme of Research in South Africa (CAPRISA) and the KwaZulu-Natal Research Institute for Tuberculosis and HIV.
To their surprise they discovered that Microsoft's algorithms for the detection of spam emails were useful in understanding the mutation of HIV.
"It turns out there are a lot of similarities between the way spammers evolve their approaches to avoid filters and the way the HIV virus is constantly mutating," a post by Steve Clayton on Microsoft Research's blog explains.
To make sense of the data the researchers hit on the idea of fine-tuning a computational biology tool, called PhyloD, with algorithms used for spam filtering. PhyloD contains an algorithm, code and visualisation tools to perform complex pattern recognition and analysis. By adding lessons learned from spam filtering it was possible to more quickly narrow in on possible areas of weakness that can be targeted for later lab research into developing therapies and possible vaccines.
Instead of trying to every possible variable and possible correlations the reapplied spam filtering algorithm created the basis of a more elegant search. Even so a huge number-crunching exercise was still needed. But access to Microsoft’s high-performance computing centre made it possible to carry out this task over a single weekend.
The work led to the discovery of six times as many possible attack points on the HIV virus than had previously been identified. Similar approaches might be applied to studies on the analysis of breast cancer and other deadly diseases, the Microsoft team reckons.
The HIV analysis research is part of a wider vaccine project. ®