"Given the similarities between the yeast and human genomes, our work should promote advances in genomics research in both yeast and humans," said Professor Timothy Hughes of U of T's Department of Medical Genetics and Microbiology, who led the research team.

A paper published in the July 9 issue of the journal Cell describes how the researchers engineered mutations to 700 of the 1,000 yeast genes that are essential to cell life. They analyzed the mutant strains by making several basic measurements -- cell size, cell shape and gene levels and by evaluating a cell's potential to grow in a variety of media. They then took these data and did computerized analysis of entire categories of genes in order to predict the functions of individual genes, applying a standard technique for pattern discovery used in fields ranging from marketing to face recognition.

"It's similar to ordering a book from Amazon.com," said Hughes. "After you've placed an order, they use the information they've gathered to predict your likes and dislikes. The next time you log onto the computer, they extrapolate and suggest other books you might enjoy. They also could use the data to predict other things for example, your age and your gender which might, on the surface, seem unrelated to books."

"We're hoping our use of this technique to predict the function of yeast genes is going to become a classical example of how to do this in biology."

To create each mutated strain, the researchers used a technique in which adding the drug doxycycline to the yeast cells disables an individual gene. This technique is a reliable alternative to the more common method of causing mutations by radiation, because the mutations a
'"/>


8-Jul-2004