TALLINN, Estonia — Where to with genome-wide association studies? What's the future path for those that study genomic variation to better understand human disease development?
Array vendors have made no secret of how they view the market: Firms such as Affymetrix and Illumina have been saying they are preparing for a second round of genome-wide association studies based on new chips with rarer variation content culled from sources like the 1000 Genomes Project. But others argue that whole-genome scans that use second-generation sequencers on smaller numbers of samples could produce more meaningful results.
Most participants at this year's Human Genome Variation and Complex Genome Analysis meeting, held here over the weekend, said they believe that array-based genome-wide association studies like those conducted over the past few years will continue in the future, but with a caveat: Factors such as study design, access to funding, and existing infrastructure will likely affect how the GWAS market develops.
Indeed, some influential scientists have been arguing that it may be more useful to take a close look at variation in a smaller number of afflicted individuals than to organize the large, expensive studies that have been popular over the past few years (see BAN 4/21/2009).
One of these is David Goldstein, director of the Center for Human Genome Variation at the Institute for Genome Sciences and Policy at Duke University. Writing in April in the New England Journal of Medicine, he said associations discovered in GWAS "reflect real biological causation," but questioned the relevance of those findings.
"Even though genome-wide association studies have worked better and faster than expected, they have not explained as much of the genetic component of many diseases and conditions as was anticipated [and researchers] … must therefore turn more sharply toward the study of rare variants," Goldstein wrote in the paper.
In it he recommended that researchers begin using second-generation sequencing to help them look at rare variation in "thoughtfully selected" individuals.
Carsten Rosenow, marketing manager for DNA analysis solutions at Illumina, said it "absolutely makes sense to look at a smaller number of individuals and see if they carry highly penetrant rare variants." At the same time, however, he said that around 5 million SNPs could be identified with every genome that is sequenced. "How are you going to identify the rare variants that actually cause the disease?" asked Rosenow, who attended the conference in Estonia.
Having enough scientific power to definitively link a marker to a disease could also be difficult, he said. "If you are going to run large-scale studies, it's going to be throughput limited. You won't be able to run in a reasonable time 1,000 or 2,000 samples to look at larger sample sets."
Eimear Kenny, a computational biologist from Rockefeller University, agreed. "I think there is still a big place for whole-genome genotyping chips," she told BioArray News here. "The feasibility of doing a whole-population, whole-genome analysis study using sequencing, while technically possible, is quite challenging in terms of storing the data and managing it," said Kenny. "Genotyping chips are here to stay."
Elaine Mardis, co-director and head of technology development at the Genome Center at Washington University School of Medicine, stressed that a technology’s ability to yield "meaningful" answers, and not its throughput or cost, should be the deciding factor in the choice of platform.
"Who cares what it costs if you don't have a tangible, meaningful answer at the end of it?" Mardis told BioArray News during the conference. Mardis, who presented data on whole-genome scans of leukemia and breast cancer samples, said her approach uses second-gen sequencing. She said the method "eliminates the biased view of genomic variation" created by using arrays and enables the evaluation of a variety of variation, such as point mutations, SNPs, CNVs, and insertions and deletions that cannot be detected on one array platform.
Mardis also cautioned that scientists who run GWAS may not be paying enough attention to the phenotypes of the samples being surveyed, which could ultimately reduce the significance of their findings.
"One of the big problems with GWAS is that for a lot of studies, people don't phenotype to the careful, narrow extent that you need to invoke the kind of power that you are talking about to find things that are definitive," Mardis said. "[Y]ou can have community standards, but, depending on the disease, those standards could be very subjective or they could be very tight."
[ pagebreak ]
According to Pui-Yan Kwok, an investigator at the Cardiovascular Research Institute at the University of California in San Francisco, limitations exist to doing additional GWAS as well as whole-genome scans.
"There will be value to [GWAS], but it will be for the people who haven't done" them he said. "For people who have already done a GWAS, it's kind of hard to redo the whole thing because the kind of money you have to put up is tough."
Kwok speculated that there will be "a lot of pressure" on array companies to keep costs equal to current platforms. "If not, I think people will be reluctant to invest in those studies," he said.
But he added that "if you use genome-wide sequencing as a matter of extended GWAS, then you need even larger samples because the amount of testing is no longer 1 million SNPs. Now you are talking about 3 billion bases.
"We cannot afford sequencing thousands and thousands of people, and we don't have enough biological knowledge to figure out what is what," said Kwok. "That being said, the good thing about sequencing is that once you have the data, the data is there. You can bank it away, and once new knowledge comes along, you can reinterpret whatever you find. So, for that, it is very valuable."
'The Cold, Brutal Reality'
For Stephen Chanock, a senior investigator at the National Cancer Institute in Bethesda, Md., despite the current debate over the future of GWAS, there has been "no quantum change" in the field.
"I am in the middle of 12 different GWA studies on different cancers in my program at the NCI at the moment," Chanock told BioArray News here. "Since 2005, that is what we have been doing. We just purchased a large number of chips and lined up the next set of studies.
According to Chanock, what those studies will look like is a "matter of when the companies improve their content and decrease their prices." But the fundamentals of the studies remain the same. "In my mind, there is no second wave," he said. "It's just one big wave that continues."
Chanock said that rounds of GWA studies that have been concluded to date have been mostly successful. "There are 500 regions of the genome and between 75 and 100 traits of diseases that we didn't know anything about" before these studies were conducted," Chanock said. "This is seeding hundreds of grants for investigators to go out and make sense of these findings."
Chanock also dismissed arguments that GWA studies had not delivered on their original promise. "The cold, brutal reality of GWA studies is that these are difficult studies that are doable, but they are just the beginning," Chanock said. "They have a tremendous requirement for follow-up and explanation. Those things don't just fall off the tree. It's not as if we have these wonderful markers that you just look up and say, 'Ah, since it falls in the coding region of this gene, we now have the explanation.' Far from it," he said.
According to Chanock, the ability to use GWA studies to comprehensively look at common genetic variation is "exciting and it gives us new clues, but it's not explaining everything and it never would have." He said that the "revisionists who say it isn't doing enough are missing the boat" because the "history of science is [the] development of better and more efficient ways to understand basic principles."
"There is the issue that GWAS were oversold by some as being able to explain more than they can," Chanock said. "That's not a consequence of GWAS; that's the PR around it. But I think that, like in anything in science, we just have to adjust our sights as we learn more."
Chanock said it was largely a financial question as to how many more GWA studies would be run on the next-gen chips being developed by Illumina and Affy. At the same time, because those arrays will contain more rare variants, Chanock said it would take GWA studies in larger populations to link rare variants to a particular disease.
"Not only is the content going to be more, but you are going to have to have studies that will be three to five times the size," Chanock said. "So, the question is, what is the market going to support for the next generation of chips?"
[ pagebreak ]
From the perspective of chip companies, customer demand for next-generation content is driving them to release new products for array-based association studies. Although Illumina and Affy, the two biggest vendors in the space, have reported a slowdown in demand for their whole-genome genotyping products in recent months, they insist it is temporary. They also have been rolling out or plan to debut new array menus for future GWA studies (see BAN 4/28/2009, 7/28/2009).
For instance, Illumina has already made its first next-gen chip available. In May, the company launched its 4-million-feature HumanOmni1-Quad BeadChip, which includes more than 100,000 new SNPs and markers identified in the 1000 Genomes Project. It also comprises more than 11,000 copy number variants from the Wellcome Trust Sanger Institute, the Hospital for Sick Children in Toronto, Harvard Medical School/Brigham and Women's Hospital, and Decode Genetics (see BAN 5/19/2009).
Affy, meantime, is gearing up to launch new whole-genome genotyping assays on its automated GeneTitan instrument in coming months. According to the firm, genotyping applications will include new content, including new copy-number variation, SNP-allele frequencies, and the on-demand sub-setting of information from an internal screen of over 1,300 individuals (see BAN 6/9/2009).
According to Illumina’s Rosenow, the firm's forecasts of a second round of GWAS are based on customer demand for access to the content being generated by 1000 Genomes and other sources.
"With the third round of the Human HapMap Project being finished now and the 1000 Genomes Project underway, people have realized that, in addition to looking at common SNPs, it's important to also look at the rare variants," Rosenow told BioArray News during the conference. "Illumina is committed to utilize that new content on their arrays and provide future products."
Rosenow acknowledged that debate exists over the best ways in which to use results produced by association studies but argued that, overall, GWA studies have been successful.
"I have been [in] the array market for over 10 years now, starting with expression," Rosenow said. "I look back on how the expression market evolved and, especially in the early years, there weren't many success stories," he said. "If I look at GWAS over the past few years, I think we have been overwhelmed by the success of the publications.
"We have more than 400 significant SNPs identified that are clearly associated with around 80 different diseases with more than 400 publications on the topic in top-tier journals," he added. "I consider that to be one of the biggest successes that we have seen in science in the past few years."
Of Illumina and Affy's next-gen chips, WashU’s Mardis said that there will be some projects that continue in the vein of existing GWA studies, especially given the inclusion of fresh content from 1000 Genomes and other sources.
"Maybe it's an 'if you build it, they will come' phenomenon," Mardis said. "There will still be some studies that carry on, and the content was in some ways designed to appeal to that," she said.
"Part of it is what you are geared up to do," Mardis added. "A lot of people have put time and effort and training into doing GWA studies," she said. "It's a fundamentally different exercise from funneling through sequencing data. Neither of them is easy. If you are doing one, it may be hard to shift gears and do another."
But to UCSF’s Kwok, the question of whether researchers will opt for array-based GWA studies or sequencing-based, whole-genome scans is second to whether there will be funds available to support these kinds of projects.
"I think the main thing is money," he told BioArray News. "In the US, all of these studies are driven by National Institutes of Health funding, and if the NIH switches its focus, it doesn't matter what people think, things will change," he said. "So, the short answer is that the future depends on funding agencies."
Echoing NCI’s Chanock, Kwok said some researchers will opt for new GWA studies based on Illumina and Affy's chips, while others, like Mardis, will decide to focus on smaller numbers of samples. Other researchers, though, may decide to undertake targeted resequencing projects based on the findings from existing rounds of GWA studies.
"There will be people sequencing samples as much as they can afford. But then the rest of us will be doing something else,” Kwok said. “Right now, that something else is following up on GWA studies and doing targeted resequencing."