Microsoft to assist in HIV research

The IT giant revealed plans to offer its data-mining algorithms, machine-learning and other software techniques as part of the ongoing research to find more effective HIV vaccines. Some of this technology is expected to be included with Microsoft’s impending SQL Server 2005 database.

Microsoft said its researchers will work with the University of Washington and the Royal Perth Hospital in Australia to identify patterns of genetic mutation in the immunodeficiency virus and in the immune systems of HIV patients. The researchers hope to use this information to propose designs for new HIV vaccines

The project is designed to illustrate how medical experts can exploit data-mining and anitspam technology to sort through millions of strains of HIV to find its genetic patterns and improve vaccines. The principle behind the research is that the software used to train anti-spam filters to recognize subtly altered words may offer suggestions as to how a vaccine can recognize and destroy HIV in the body.

Simon Mallal, executive director of the Centre for Clinical Immunology and Biomedical Statistics at Royal Perth Hospital and Murdoch University, said: “Microsoft Research’s contributions enabled us to filter patient data 10 times faster than any previous research technique we’ve used and produced vital clues about the building blocks of a vaccine – clues that were all but impossible to find in our growing stockpile of medical data.”

The announcement comes after an initial project was set up to develop a preliminary vaccine “design” using Windows systems. The work has now branched into three research initiatives, one of them focusing on creating graphical models to investigate the mutation of HIV and how it is influenced by a person’s immune system. These graphical modeling capabilities will ship with Microsoft’s SQL Server 2005.

The other two initiatives involve software yet to be seen in Microsoft’s line of products. One uses machine learning to study reactions of the immune system to foreign proteins found on infected cells, exploiting a technique similar to spam-filtering technology, which will enable algorithms to separate good and bad cells.

The last area of research will apply an algorithm called Epitome, which was originally developed to compress images, to compress foreign proteins to develop better vaccines.

New designs are being tested at the University of Washington using samples taken from HIV-infected patients. Additional tests are scheduled at the Royal Perth Hospital.

Microsoft said that results from both trials should be available later this year.