Biotech--J. Gillis, The Post: "Glass Chips as a Window on Disease"
washingtonpost.com
>>> Glass Chips as a Window on Disease
By Justin Gillis
Washington Post Staff Writer
Tuesday , October 3, 2000 ; Page E01
A California company is launching a new venture to put microscopic detectors capable of reading the entire human genome onto a set of glass chips, then use them to detect common genetic variations in the human race and to match those patterns with disease.
The plan, devised by Affymetrix Inc. of Santa Clara, Calif., is to scan the complete genomes of at least 50 people next year, with the goal of producing the most exhaustive human gene database ever created.
With that database, scientists hope to zero in on key genetic patterns to produce a smaller set of "gene chips"--then use those to scan thousands of people with heart ailments, cancer, diabetes and the like to see which genetic variations correlate with what diseases. The idea is to generate, more rapidly than in the past, new theories of how to treat disease that pharmaceutical companies could use in creating medicines.
The plan is to be announced to investors and Wall Street analysts this morning. Affymetrix is one of the pioneer names of the new biology, widely respected as the company that invented gene chips.
Affymetrix has lured a leading geneticist, David Cox of Stanford University, to be scientific director of the new venture, which will be a spinoff company called Perlegen Sciences Inc. The idea, Cox said in an interview, is to try to break through the frustratingly slow pace at which genetic research is yielding new medical treatments.
"My whole professional career has been spent trying to figure out how genetic information can be used to help people," said Cox, who is 53. "I could make piecemeal contributions with the present technology, but I'll be dead before I know whether it works. My motivation is that I want to be alive to know if it works. I've spent 30 years of my life with the hope that something like this would come along."
The Affymetrix plan builds on the dramatic announcement at the White House earlier this year that competing groups of scientists had created the first maps of the human genome, the entire set of genetic instructions that encode the biology of the human race.
But those gene maps are "consensus" maps--essentially a sort of average, the genetic code of some hypothetical Everyman. The great, daunting task before biologists is to use that average map, containing about 3.1 billion units of information, to begin to understand how people diverge from the average--and how those divergent sets of genes interact with how people live or what they eat to make them susceptible to various ailments.
Researchers are already trying to do that, of course, and have made notable headway in some diseases. But the existing technologies--which include earlier generations of gene chips from Affymetrix--are still slow and expensive when measured against the scale of the task.
The new technology will work this way:
Affymetrix and Perlegen will etch 5-by-5-inch glass wafers to create millions of microscopic pillars, in much the way the semiconductor industry etches transistors onto silicon. Atop the pillars, scientists will deposit tiny strands of genetic material of known composition to create molecular structures called genetic probes.
DNA--deoxyribonucleic acid, the carrier of hereditary information--will be extracted from human volunteers, multiplied billions of times by chemical methods and then washed over the wafers. A stretch of DNA floating by a microscopic probe of precisely matching composition will bind to it like a magnet. Fluorescent dyes added to the mixture will reveal spots where DNA has attached to a probe, revealing the exact composition of that stretch of DNA in the body of the person from which it came. The fluorescent dyes will be read by special scanners and turned into data.
Cox said mathematical analyses suggest that scanning the genomes of 50 people should reveal the existence of any genetic variations present in more than one out of 10 people--enough, he believes, to come up with reliable correlations between genetic patterns and common diseases.
If the technique works, its power will come from the fact that genes do not move down through the generations entirely at random but instead are inherited in large groups that stick together. This presumably explains, as an example, why people in some families tend to die of heart disease while those in other families get cancer.
Cox said the initial scan of 50 people should reveal which gene variants are being inherited in groups. With that information, Perlegen scientists hope to boil the genome down to a much smaller set of gene chips able to detect particular patterns of inheritance.
With this smaller chip set, they will be able to run tests on thousands of people in all stages of health, correlating the genetic results with medical information. Perlegen--from the Latin perlego, to scan thoroughly--expects to strike deals with drug companies to look intensively at particular diseases.
The ultimate goal is to tell the drug companies what genes they need to affect as they design new pharmaceutical compounds. Perlegen would most likely make its money by receiving multimillion-dollar payments from drug companies as it hits critical milestones.
There is no hard evidence to prove the venture will work--failures or bottlenecks at any of several complicated steps could derail it. Moreover, for various technical reasons, it's possible Perlegen will wind up creating a "noisy" database with a lot of bad information, which would severely limit its usefulness. But Stephen P.A. Fodor, chairman and chief executive of Affymetrix, said preliminary research has convinced him that the approach is workable.
Executives at Affymetrix have been cooking up the Perlegen plan in secrecy for much of the past year under the initials GSP, for Genome Scanning Project. They have already hired a chief executive, Brad Margus. To get the company off the ground, they are shooting to raise about $100 million in private capital. No public offering of shares is planned in the near term, though that's a possibility down the road.
© 2000 The Washington Post<<< |
