Perlegen Sciences' CEO Brad Margus with his sons Quinn (left) and Jarrett (right).

Few people would envision a former Florida shrimp processor leading a multimillion-dollar Silicon Valley biotechnology company. But Brad Margus is no ordinary shrimp processor. Margus, the CEO of Perlegen Sciences in Mountain View, Calif., has two sons with ataxia-telangiectasia (A-T), a rare genetic disorder that causes neurological decline and early death, most often in the mid-20s. In 1993, when his toddler sons were diagnosed, Margus was a Harvard MBA heading up a south Florida company that sold shrimp to restaurants, cruise lines, and hotels.

After the boys' diagnosis, Margus and his wife Vicki founded the nonprofit A-T Children's Project, and in seven years raised $7 million to help identify the gene that causes the disease, which affects about 500 children in the United States. In 2000, Stanford geneticist David R. Cox was working with Affymetrix CEO Steve...


Last October, the company won a $6 million contract from the National Human Genome Research Institute to add 2.25 million additional SNPs to the International HapMap Project. The public-private partnership was launched in 2002 to map the 0.01% of DNA responsible for human genetic variation by analyzing the genomes of 270 people from China, Japan, Nigeria, and the US. The international team achieved its original goal of completing a first draft of the human haplotype map, which contains 1 million SNPs ahead of schedule. With Perlegen, the consortium now plans to test another 4.6 million SNPs in public databases and add them to the map by the end of the year.

"They developed an outstanding application to the request for applications that we put out for cheap and accurate genotyping for the HapMap," says Lisa Brooks, program director for the National Human Genome Research Institute. "They bring a deep scientific understanding to the project, and they are easy to work with."

Perlegen's quest is to discover genetic variants associated with disease for potential new therapeutics and diagnostics. The researchers are focusing on Alzheimer disease, autism, breast and colorectal cancers, diabetes, metabolic syndrome, nicotine addiction, Parkinson disease, and rheumatoid arthritis. The company has also announced genetic variation studies in lab mouse strains and Salmonella.

In total, Perlegen has partnered with eight different NIH institutes on projects worth more than $20 million. In March, Perlegen won a grant from the US National Cancer Institute to study tumor-specific DNA mutations involved in colorectal cancer with the Sidney Kimmel Comprehensive Cancer Center at Johns Hopkins. In April, it won a grant from the National Institute of Arthritis and Musculoskeletal and Skin Diseases to conduct a high-density whole genome association study of rheumatoid arthritis with DNA samples from patients, and expertise provided by the North American Rheumatoid Arthritis Consortium.

And in May 2005, Perlegen won a biodefense-related grant from the National Institute of Allergy and Infectious Diseases to analyze the genomes of 150 strains of plague-causing Yersinia pestis and Y. pseudotuberculosis.


Perlegen's competitors include Applera Corporation, Genaissance Pharmaceuticals, Sequenom, and others. The company, which has 115 employees, also has offices in McLean, Va., and Tokyo, Japan. Affymetrix owns more than 25% of the private company, and Perlegen has not disclosed information regarding a potential initial public offering. However, it recently completed a private financing to help with its expansion and in-licensing efforts. It was one of the larger private life sciences financings for the first quarter of 2005 with $74 million from Tokyo-based CSK Venture Capital Co. Ltd. At a time when venture capital funding for some biotech companies can be a struggle, funding for Perlegen has been robust. Perlegen has locked up $206 million in venture funding to date.

"There were a surplus of venture capitalists wanting to fund [genomic research], but there was a shortage of seasoned business people who understood genetics," Margus says. The company's revenues quadrupled in one year – from $7 million in 2003 to a little more than $28 million in 2004.

However, the company does have stiff competition in its sector. "I think that Perlegen has a very successful genotyping platform, [but] it is entirely unclear what company, if any, has a clear lead at this time," says David Altshuler, of the Broad Institute of Harvard University and Massachusetts Institute of Technology. "Multiple companies – Perlegen, Affymetrix, Illumina, ParAllele, Sequenom – offer exciting technologies... and the leading companies one year are often replaced by other companies with a new technology the next. The overall market size and composition is unclear as well," Altshuler adds.

"Perlegen is not the untouchable gold standard in genotyping," agrees Andrew Clark, of Cornell University. "They have a very good, high-throughput platform that works well for their problems, but hybridization methods fail to get SNPs in regions of redundancy and in regions with particular sequence motifs. They lose accuracy in regions of extreme base composition.... It is not very flexible, so that an experimenter can change the set of SNPs they want to genotype without manufacturing another expensive array."


Part of Perlegen's business strategy is to partner with Big Pharma – such as GlaxoSmithKline, Bristol-Myers Squibb, Eli Lilly, and Johnson & Johnson – to analyze SNPs in patient populations. The goal is to correlate genetic variants with therapeutic responses to a drug, which could then be used to improve drug efficacy, reduce adverse side effects, and improve the market position of drugs. This data could also be submitted to the US Food and Drug Administration to help in the approval process.

"Pharma companies may invest hundreds of millions of dollars in a new drug's development, only to find that while it is very efficacious in some patients, it does not help others," says Perlegen spokesman Paul Cusenza. "However, it may be possible to find genetic markers that are highly predictive of those with the outstanding efficacy."

Perlegen conducts high-density whole genome association studies in pharmacogenomics for their partners. This results in the identification of a set of associated genetic variants – typically ranging from 20 to 40 SNPs – which help predict those patients who may have an adverse event or insufficient efficacy when given a certain drug. For example, if a patient has more than 20 predisposing alleles within a set of associated SNPs, then they have at least a 14.2 relative risk for an adverse event. Eliminating those patients from the treatment group could cut the number of side effects by 40%. In this example, however, 3% of patients eliminated from a study would be false positives, meaning they could have safely taken the drug without adverse effects.

The company is also looking to license drugs that have been dropped by other companies to see if it can identify certain patient populations that may benefit from the drug.

In January 2004, Perlegen announced a research pact with AstraZeneca of London to discover gene variants that predispose patients to heart attacks, and two deals with New York-based Pfizer: one to study drug responses in patients with major depression disorder, and the other to focus on research to better understand metabolic syndrome, with the goal of sharing in the therapeutic and diagnostic rights resulting from the collaboration. In April, Perlegen and Mitsubishi Pharma Corporation announced a licensing agreement that gives Perlegen exclusive worldwide rights, outside of Asia, to develop and commercialize customized treatments for diabetes and other metabolic disorders.

Late last year, Perlegen won a $2.8 million grant from the Michael J. Fox Foundation for Parkinson's Research to work with the Mayo Clinic in Rochester, Minn., to determine the SNPs related to Parkinson disease.

"No one else can do it," says Demetrius Maraganore, of Mayo's Department of Neurology. "Perlegen has unique knowledge of the human genome structure and of variations in the human genome." And this past February, Perlegen announced it was teaming with the University of Cambridge, Cancer Research Technology, and Cancer Research UK to conduct a high-resolution, whole genome association study on breast cancer.

Margus acknowledges that nothing his company has done so far has helped to cure his sons' disorder. Jarrett, now 16, and Quinn, now 14, have developed tremors, garbled speech, and problems with muscle coordination. Margus forges ahead with research he hopes will find cures for other diseases. "It's always the same issue when you make a discovery," Margus says. "How do you apply it in the clinics so it makes a difference in people's lives? I wanted something that happened in my lifetime, or my kids' lifetime – not some day." But, he adds: "Life is short. This is an exciting place to be if you appreciate what sorts of answers could be."

Interested in reading more?

Magaizne Cover

Become a Member of

Receive full access to digital editions of The Scientist, as well as TS Digest, feature stories, more than 35 years of archives, and much more!
Already a member?