Speeding the search for new human genes.

A group of genetics researchers has proposed a new way of tackling the Human Genome Project, the $3 billion, 15-year federal initiative to decipher every gene in the blueprint for human life. The new scheme, they say, could shave several years and millions of dollars off the project, which began about a year ago.

The team, led by J. Craig Venter at the National Institute of Neurological Disorders and Stroke in Bethesda, Md., sketched out a rough picture of part of the human genome by identifying segments of genes that are turned on in the brain. Venter says such bits of genes -- called expressed sequence tags--could serve as dots outlining the genome, providing researchers with starting points for locating and deciphering the code for all of the estimated 50,000 to 100,000 human genes.

Researchers in Venter's laboratory, together with geneticists at St. Elizabeth's Hospital in Washington, D.C., demonstrated their approach using molecular "libraries" of all of the genes actively functioning in the human brain. They chose to focus on the brain because its functioning requires 30,000 genes, 20,000 of which are thought to be used by no other body tissues.

Venter's group randomly selected and deciphered 609 short stretches of DNA from the molecular libraries. Because the libraries had been back-translated from messenger RNA -- the chemical intermediate DNA uses to make proteins -- these researchers knew they were working with genes that code for proteins, and not with the regulatory sequences of DNA or the "filler" segments that together constitute 97 percent of the human genome. The team reports in the June 21 SCIENCE that most of the deciphered bits matched known genes, but 230 of the tags identified genes never before discovered.

"We tried this approach to see if just a little bit of sequence from each gene would be enough to identify it . . . and it worked," Venter says.

He estimates that the technique could identify and decipher genes up to 1,000 times faster than the current approach, which requires geneticists to read the sequence of roughly 1,000 overlapping stretches of DNA to identify a single gene. With the new method, Venter says, "we can get partial sequences from 1,000 genes" with the same amount of effort. Moreover, because the partial sequences need to be read only once -- instead of the multiple times now required -- he calculates that the approach could slash the cost of reading one unit of DNA from the present $1 to $4 to just 12 to 15 cents.

In August, Venter will receive a $1 million grant from the Department of Energy (DOE) to apply the new technique to the Human Genome Project, which is jointly sponsored by DOE and the National Institutes of Health (NIH). He says he hopes to identify all of the genes functioning in the human brain within the next three to five years. If the technique is adopted by other gene-mapping labs around the world, all human genes can be identified by 1996, Venter predicts.

Benjamin Barnhart, who manages DOE's human genome activities, says his agency supports Venter's strategy because it "gives us a handle on the functinal portion of the human genome." Barnhart expects the panel of outside experts who advise DOE and NIH on the Human Genome Project to discuss the proposal when they meet next week. At the same time, he stresses that the genome project's ultimate aim is to decipher every human DNA sequence, not just to locate protein-encoding genes. Venter's approach "hasn't in any way detoured us from the long-term goal of providing the scientific community with the sequence of the entire genome," Barnhart says. The DNA sequences interspered among protein-encoding genes in the human genome could turn out to have surprising functions, he contends.

Elke Jordan, deputy director of the NIH's National Center for Human Genome Research, remains noncommittal about the new strategy. She says Venter's group has essentially refined an approach outlined a year ago by Maynard Oson at Washington University random unique DNA sequences -- not just those that are part of genes -- to expedite the Human Genome Prokect. When asked whether NIH will adopt Venter's technique, Jordan ever approaches make sense."