science

A refined map of the human genome shows that humans have even fewer genes than previously thought -- less than 25,000, about the same as a mustard green.

Scientists originally guessed that humans might have more than 100,000 genes. A government-funded project launched in 1990, the Human Genome Project, and Celera, a privately funded company, raced neck and neck in a competitive and sometimes contentious contest to find the true number of human genes first.

The human gene count decreased to about 30,000 in 2001, and the new, lower number has researchers wondering how the human body operates on such a low gene budget.

"It's astounding that we get by with so few protein-coding genes, but that seems to be sufficient because here we all are," said Francis Collins, director of the National Human Genome Research Institute.

Researchers over-counted the number of human genes in the past, Collins said, because the human genome has many DNA regions that look like they could be genes, but are actually defunct. Newer technologies and closer study separated the live genes from the dead ones.

The Human Genome Project and Celera each claimed their gene-sequencing techniques were superior. But they called a truce, and in February 2001 at the White House, Collins and Craig Venter, then-president of Celera, announced together they had each completed drafts of the human genome sequence.

But three years later, new research not only revises the number of human genes, but also settles the battle over sequencing methods. Venter developed Celera's method of decoding genes, called "whole genome shotgun," in 1996. The new research doesn't reject the shotgun method outright, but says that for more difficult parts of the genome -- areas containing duplicate strands of DNA -- it will deliver inaccurate results and a backup method is necessary.

"Large duplications and genes that are embedded within them will be lost" because algorithms used in the shotgun method can't distinguish virtually identical sequences from each other, said Evan Eichler, an associate professor of genome sciences at the University of Washington in Seattle, and lead author of the paper that analyzes the shotgun technique. A separate paper outlines the refined map. Both studies were published in the Oct. 21 issue of Nature.

However, for parts of a genome that don't contain repetitious DNA, the shotgun method should work just fine. Repetitious DNA accounts for only 5 percent of the human genome, and the problem can be solved by following up the shotgun approach with the more traditional method (known as BAC, or bacterial artificial chromosome).

"If you want a very accurate sequence, it's pretty clear you can't stop with shotgun," Collins said. "But if you want lots of information in a short period of time, the shotgun will give that to you, you just can't expect it to be complete."

Some genome researchers who adopted the shotgun method will now know the best way to get a fine-tuned, complete genome map.

"There may be a number of small problems left, some due to the whole genome shotgun method, which might be solved in a later phase of the project," said Hugues Roest Crollius, a researcher at Ecole Normale Supérieure in Paris, who published the sequence of the Tetraodon puffer fish genome, also in the Oct. 21 issue of Nature. "The main advantage of this method, however, is that we reached the current stage much faster and at much lower cost than using conventional methods."

That was the point of the analysis of the whole genome shotgun method, said Granger Sutton, a senior computational scientist at the J. Craig Venter Institute and a former Celera executive who participated in the shotgun study.

"What's nice about the paper is it's a clarification of to what degree the problems exist and specifically what the (difficult) repeats will look like," Sutton said. "They will be longer than 150,000 base pairs and more than 97 percent identical."

The shotgun technique involves blasting apart the genome, sequencing random fragments and then piecing them together using computer algorithms. Using the BAC-based technique, scientists sequence sections in order, placing them physically on a map.

In 2001, both the Celera and the Human Genome Project maps were missing big chunks of sequence and some parts were misassembled. The updated version contains almost 3 billion letters, or bases (A, C, T and G, which represent the nucleotides that make up all DNA: adenine, cytosine, thymine and guanine), with only 341 gaps (the previous version has 150,000 gaps) and an error rate of one per 100,000 bases.

Celera published (.pdf) one update on its map in late 2001, but the company later switched its focus to drug development and abandoned gene-sequencing efforts.

The next step will be for geneticists to begin deciphering the remaining 1 percent of the genome, said Lincoln Stein in an article that accompanies the human genome paper. The technologies to do so haven't yet been invented.

"We ultimately hope that somebody will invent new ways to detect those sequences," Collins said.