When the first American strain of the AIDS virus was isolated, researchers at the National Cancer Institute needed to understand its genetic details quickly.

To get that information, NCI hired a team of scientists at E.I. Du Pont de Nemours & Co. in 1984 to run the routine but laborious procedures needed to determine the order of the virus's genetic alphabet, a sequence of some 9,000 A's, T's, G's and C's. The arrangement of the genetic alphabet determines the shapes of the virus's proteins and enzymes, controls how the virus behaves in an infected cell and determines its deadliness.

The crash project took six months. Today, it could be done in a day.

Automation and high-tech engineering has come to molecular biology. Where it once took a researcher and technicians months or even years to determine the genetic sequence of even the smallest genes, new gene sequencing machines have compressed that time into days or weeks through the use of high speed computers and lasers, and both fancy and traditional biochemistry. Not only are the gene machines faster, but they promise to be more accurate.

In a recent issue of Science magazine, Du Pont announced the development of a new sequencer -- the Genesis 2000 DNA Analysis System -- that uses novel fluorescent dyes and a new detection system to perform in six hours a study that used to take weeks.

Applied Biosystems Inc. of Foster City, Calif., already has a similar gene sequencing machine on the market, and two more reportedly are coming from the European Molecular Biology Organization (EMBO) and EG&G Biomolecular of Watertown, Mass. The Japanese, too, have been working on a gene sequencing system that uses robots to automate nearly the entire process.

These new machines "will have a profound impact {on research}, which is extremely difficult to foresee," said Dr. Norman Anderson, chief scientist of Large Scale Biology Corp. in Rockville, yet another firm involved in this new technology.

Dr. LeRoy Hood, chairman of biology at the California Institute of Technology and leader of the first group to make an automatic DNA sequencing machine, agreed, saying, "I think there will be a spate of DNA sequencers that will come on the market. There will be a lot of intense competition to design a good machine. Researchers will be able to shop around and pick the machine that will do the best job. It is all to the research community's enormous advantage. The difficulty will be sorting out the claims that the companies will make."

While the machines themselves are important advances in technology, the biggest impact could be on a supersized biology project often referred to as the human genome project. For the last two to three years, the biological community has been debating the wisdom, value and cost of determining the sequence of all 3.5 billion letters that make up the entire complement of genes in the human body.

Some scientists argue that the information gained would have a profound impact on medicine, agriculture and evolutionary biology, while opponents argue that the $3 billion to $6 billion the project is currently estimated to cost spent over 10 to 20 years would simply divert too many resources away from other important research.

The new, more economical gene machines, however, could begin to tip the balance in favor of the project.

Congress already is paying attention to the proposed genome project, with the House authorizing some $30 million and the Senate authorizing $6 million in the National Institutes of Health's budget for this fiscal year. Whether any of that funding will survive the new emphasis on reducing the federal defecit is not clear, but the next several weeks are expected to be critical.

The Department of Energy, too, has become a player in the genome project, with several million dollars already included in next year's budget for the development of sequencing technology.

"The sequencing project really has been reaffirmed by meetings going on this week," Dr. James B. Wyngaarden, director of the National Institutes of Health said in a recent interview. Leading scientists, including Caltech's Hood and representatives from Du Pont, came to the NIH recently to describe progress in the field.

Nearly a year ago, Wyngaarden's advisory committee laid out the technical advances required for a mammoth human genome project, including, among other things, improvements in computerized handling of the massive amount of data expected from such a project, speeding up existing gene mapping projects, and advances in technology to speed up the sequencing process and to reduce its cost.

The machines "are still in the developmental phase," Wyngaarden said. But "the cost is coming down rapidly. Without these machines, the sequencing cost is about $1 per nucleotide {one genetic letter}. The companies are pretty sure the cost can be brought down to 3 or 4 cents per nucleotide."

The cost improvement means the difference between a $6 billion project over 10 years and a $300 million to $600 million project over 10 years, Wyngaarden said. "Those sums are not so staggering. At $30 million a year, we would be able to handle it as a priority item within the {NIH's $6 billion a year} budget. It clearly is not in the $30 million range yet, but the estimate is that in five years, it will be. Maybe sooner."

The recent advances began when Hood's group figured out how to modify and automate one of the two methods for sequencing genes. Instead of marking the genetic material with radioactive tracers, Hood's group used off-the-shelf fluorescent dyes to mark fragments of DNA. A different dye is used to identify each letter. The fragments are then separated by size with gel electrophoresis -- a kind of jellylike sieve. A laser identifies each of the fluorescent dyes as they move through the gel.

Although 75 machines have been sold by Applied Biosystems Inc., there reportedly have been technical difficulties, especially with the dyes, but Hood says they are being worked out.

The Du Pont effort also replaced radioactive labels with laser-fluorescent dyes, but through a fundamentally different approach.

"It was clear we could not do a 'me too' approach," said James M. Prober, a researcher at Du Pont's experimental station. "The Du Pont bureaucracy would not let it happen. It quickly became clear we needed some classy chemistry. We would have to come up with something novel."

The something novel was a brand new set of dyes that could be hooked to the end of a DNA fragment, instead of the beginning as does the ABI machine. The difference sounds trivial, but it has both biological and practical significance.

For example, instead of having to run a sequencing reaction four times -- once for each genetic letter -- the Du Pont machine can run the reaction once, with four different dyes marking the four different genetic letters at the same time. That alone speeds up the procedure four times.

For a huge program like the human genome project, speed is essential. The Du Pont machine performs a run in five to six hours, compared with 12 hours for the ABI machine. But the ABI machine runs 16 samples at one time; Dupont runs only 12. Working at full capacity, both machines should be able to identify 10,000 to 15,000 genetic letters each day.

Accuracy, however, is as important as speed. "Fluorescence and lasers give us very high sensitivity," said Du Pont's Prober. "It is like detecting a grain of sand in Niagara Falls."

The laser's high sensitivity provides an accuracy rate estimated at around 1 percent for both the Du Pont machine and the ABI machine. "My feeling is it is going to be substantially more accurate than the current {hand} methods," Prober said.

The basic Du Pont machine, including the Apple MacIntosh computer that runs it, is expected to cost around $90,000, said Karl H. Jahn, Du Pont program manager for the sequencer. The price of the ABI machine is comparable. With some 200,000 scientists sequencing genes, Du Pont predicts at least a $100-million-a-year market for sequencing machines in the United States alone, with the Du Pont machine initially capturing perhaps 20 percent of the market.

The market could grow because the technology will make previously daunting sequencing tasks seem reasonable. Until now, some scientists in small laboratories have left the complicated sequencing work to larger labs. "As soon as you can do 5,000 to 10,000 {genetic letters} per day, you will think about it," Jahn said. "They would be able to do tasks that they would not be able to undertake before."