|In yeast, tackling the problem of too many unknown genes|
By Kate Dalke
July 19, 2002
With new genomes being sequenced all the time, researchers are discovering thousands of previously unknown genes. More than ever before, scientists need efficient methods to determine the biological functions of these genes. In a new study of yeast, researchers have developed a computational method to predict the function of nearly 1,650 yeast genes.
"As one genome after another gets sequenced, we're still overlooking the functions of a majority of genes," says Timothy R. Hughes, who co-led the study at Rosetta Inpharmatics in Kirkland, Washington. "It's not surprising that there is a bunch of 'leftover' genes," he adds, "because there is really no one out there looking for them."
To address the problem in yeast, Hughes teamed up with Rosetta colleagues Steven J. Altschuler and Lani F. Wu. They focused on some 2,000 genes (a third of the S. cerevisiae genome) that still have no known biological functionsix years after the organism was sequenced. The findings are published in Nature Genetics.
The researchers identified, for example, 285 genes involved in the processing of RNA, which is one of the least understood functions in cells. To evaluate their predictions, the researchers tested some genes experimentally. Now at the University of Toronto, Canada, Hughes is investigating those genes involved in RNA processing.
To pursue their quarry for genes, the researchers created a computer program that combines different algorithms to identify yeast genes with similar expression patterns and therefore similar functions. The goal was to link unknown genes to genes that have already been placed in a functional category.
"Trying to find an algorithm to group the function of 6,000 genes is hard," notes Altschuler who, together with Wu, developed the computer program to sort data into clusters of similarly expressed genes. "You have to tell a computer what you mean by grouping genes together. And there is a certain kind of art, math and statistics in doing that," he adds.
One useful feature of the method is that each gene can be assigned to more than one functional category. This reflects what actually happens inside cells, where genes and proteins may perform a variety of jobs.
Altschuler and Wu, now at the Bauer Center for Genomics Research at Harvard University in Cambridge, Massachusetts, are currently studying how groups of proteins work together and trying to identify principles that characterize their behavior.
See related GNN article
. . .