Michael Brent, Ph.D., Washington University professor of computer science and engineering, used his unique software, TWINSCAN, on the genome of Caenorhabditis elegans (C. elegans). The genome of another nematode C. briggsae, was also used to determine which parts of the sequence have changed since the nearest common ancestor of the two species. He found first of all that TWINSCAN predicted 60 percent of the genes in the C. elegans genome exactly, right own to the last amino acid.
"This (60 percent) is a new level of accuracy for a complex genome," Brent said. "It's quite a step up from what you see in the human genome, for instance, where not even a third of the genome can be predicted exactly. The 60 percent is the highest accuracy published for a multicellular organism."
C. elegans is a biological model for animal development and genetics, and is the first animal genome to be sequenced, back in 1998. Nematode researchers rely on a genome annotation database called WormBase. Along with confirmed genes, WormBase includes thousands of predicted genes without evidence from complementary DNA (cDNA) or expressed sequence tags (EST), which help locate genes. These predicted genes are derived by a combination of a program from the previous generation and some curation by human experts. Brent and his colleagues say that the accuracy of WormBase can be improved with the use of TWINSCAN predictions. And Brent predicts that the age of the human genome annotator is passing ?the future belongs to computer-driven annotation.
Crossing the tipping point
"We've crossed the tipping point with gene prediction where it's
Source:Washington University in St.Louis