Featured Article
Monday October 19, 2009
by Jeffrey M. Perkel
Genetics moves at a dizzying pace these days. Research journals trumpet new genetic associations almost daily for everything from disease predisposition to drug responses to physical traits. As of Oct. 6, 2009, the National Human Genome Research Institute's Catalog of Published Genome-Wide Association Studies (GWAS) lists 416 such studies dating back to 2005. More focused studies over the same period surely number in the thousands.
The genetic markers at the heart of these publications are single nucleotide polymorphisms, or SNPs. A SNP is an inherited, single-base variation between two individuals at a particular chromosomal position. On average, humans contain one SNP every couple hundred bases, and millions have been indexed in dbSNP; millions more are pouring out of massive sequencing efforts like the 1,000 Genomes Project.
Consider, for instance, the polymorphism called rs10993994. Located at position 51,219,502 on the long arm of human chromosome 10, rs10993994 is a C/T polymorphism: Some individuals possess a C at this position; others contain a T. Usually, the consequence of having one variant or another is negligible; SNPs serve mostly as genetic landmarks. In this case, though, the consequences are very real: the polymorphism is associated with a higher risk of prostate cancer, and according to a study published earlier this year in the Proceedings of the National Academy of Sciences, that may be because the T variant dampens expression of a nearby gene, MSMB.1
Stephen Chanock, chief of the Laboratory of Translational Genomics and director of the Core Genotyping Facility at the National Cancer Institute, who led the study, says researchers need to carefully consider both their experimental design and available resources before diving into genotyping. In particular, he stresses the importance of patient cohort size.
"You need to think very carefully about power calculations," he says. "One of the daunting challenges is to conduct a GWAS that is adequately powered to detect the estimated effect size."
In Chanock's case, power was not an issue. His study followed a familiar pattern in modern genetics: start by casting a wide net on relatively few samples, and then focus in on selected regions using larger sample numbers. He screened nearly 528,000 polymorphisms in each of 1,172 prostate cancer patients and 1,157 controls, narrowed his focus to 27,000 SNPs in 7,900 individuals, and finally homed in on 13 SNPs in a total of 12,200 samples.
To run these genotypes, Chanock's group used two technologies: BeadChip microarrays from Illumina, and TaqMan® assays from Life Technologies. The former, which Chanock used for the genome- and mid-level analyses, are available off-the-shelf in configurations containing from hundreds of thousands to more than one million polymorphisms per array (for instance, the Human Omni1-Quad, with four arrays per chip), and as custom iSelect arrays for probing from 3,000 to 200,000 polymorphisms in each of 12 samples per chip. Both formats run the Infinium II assay, a combination DNA hybridization and single-base primer-extension reaction that interrogates each SNP based on which fluorescently labeled nucleotide is added.
Life Technologies' Applied Biosystems TaqMan® SNP Genotyping Assays are PCR-based assays in which passage of the DNA polymerase during the extension reaction cleaves a fluorescently labeled (but quenched) oligonucleotide probe situated over the polymorphism, releasing the fluorescence signal indicating which alleles are present in the sample. Classically, these TaqMan® assays are single tube assays, but you can also use Life Technologies' custom plating service (in either 96- or 384-well formats) for TaqMan® assays. (The company offers some 4.5 million validated TaqMan® genotyping assays from which to choose, or you can supply your own.)
Recent developments have boosted the throughput of TaqMan® assays. Life Technologies' Applied Biosystems™ TaqMan® OpenArray® Genotyping System, for instance, enables researchers to run 3,072 33-nl reactions in parallel on a single array.
"You can run up to 4,600 samples without robotics with a single person in a day," says Life Technologies genomic assays product manager, Elizabeth Goley. "That's 32 plates, or 98,304 genotyping reactions, per day."
Alternatively, the Fluidigm Dynamic Array for Genomic Profiling uses microfluidics and automation to run 48 or 96 TaqMan® assays on each of 48 or 96 samples (called 48.48 and 96.96 arrays, respectively). Or, you can order a custom TaqMan® plate (in either 96- or 384-well formats) directly from Life Technologies.
Jiannis Ragoussis, head of genomic research at the Wellcome Trust Centre for Human Genetics at the University of Oxford, who authored an overview of genotyping technologies in the September Annual Reviews of Genomics and Human Genetics2, has also used whole-genome microarrays from Illumina in his research, as well as a second Illumina technology called GoldenGate, Affymetrix GeneChips, and iPLEX from Sequenom.
Each, he says, targeted a different SNP density. For 12 to 400 SNPs per sample, his team used iPLEX, a mass spectrometric approach that calls SNPs based on the mass of the products of single-base primer extension reactions, and which can be multiplexed up to 40 SNPs per reaction. According to Gianfranco de Feo, senior director for marketing at Sequenom, the company recently released a new version of their MassArray mass spectrometer—the MassArray Compact 96—to accommodate 96-well-formated plates (the original version used a 384-well format). "Now low-to-medium-throughput labs can take advantage of the flexibility, high quality data, and low per-reaction costs characteristic of the MassArray system, at an even lower system cost," de Feo says.
For 400 to 1,536 SNPs per sample, Ragoussis used GoldenGate, a combination primer extension reaction, oligonucleotide ligation, and PCR, which calls a SNP based on which primer is able to initiate primer extension (and thus, to participate in the final PCR step).
Finally, for higher SNP densities, his team used Illumina's Infinium II BeadChip arrays or Affymetrix GeneChip® microarrays (the latter when Illumina had no comparable off-the-shelf product—for instance, when doing certain mouse SNP studies—or when focusing on copy-number variant (CNV) coverage).
"The Affymetrix chip contains more assays [per array] and provides a better CNV region coverage," he says.
GeneChip® microarrays call genetic variants based on selective hybridization to perfect match (as opposed to mismatch) primers; the company's current flagship, the Genome-Wide Human SNP Array 6.0, probes 1.8 million SNPs and CNVs. Traditionally, Affymetrix has manufactured arrays like the SNP Array 6.0 in a single-sample cartridge format. But, in response to feedback indicating customers needed to be able to process more samples faster and more easily, the company recently announced the new Axiom™ Genotyping Solution.
The system's consumable is the Axiom Array Plate, containing about 700,000 markers on each of 24 or 96 arrays in a microtiter plate format. Using up-front automation (Beckman Coulter’s Target Prep Express), a new genotyping assay, and the new GeneTitan™ Multi-Channel Instrument, "Customers can run eight plates [up to 768 samples] per week with minimal user intervention and hands-on time," says Jay Kaufman, vice president of product marketing. "It's considerably higher throughput than anything else on the market."
The Axiom Solution uses a different genotyping approach than did previous GeneChip microarrays, Kaufman says. Those earlier arrays (including the SNP Array 6.0) were one-color assays based on the so-called "whole-genome sampling assay," which is constrained by the locations of restriction sizes within the genome. "Because of that, you are somewhat limited in the ability to access any portion or SNP in the genome," Kaufman says. In contrast, Axiom is a two-color, "random-access assay. This gives you greatly improved access to SNPs on the genome."
At genetic research and service firm deCODE Genetics, researchers also use Illumina whole-genome microarrays (including off-the-shelf 610-Quad, 660-Quad, and 1M configurations), and have applied them to more than 100,000 individuals, estimates Richard Leach, the company's vice president of scientific services. However, once they identify a genetic "hot spot" in first-pass GWAS, they switch to a lower-plex option: the Nanogen Centaurus assay.
"We will find every sequence variation we can and go after it," says Leach. "We call that 'carpet bombing.'"
Centaurus is an assay that takes advantage of a DNA repair mechanism. It uses a pair of primers flanking the SNP and a coupled (quenched) dye to mimic the abasic DNA sites that are the targets of the enzyme, endonuclease IV. Successful hybridization of oligos both 5' and 3' of the SNP gap induces the endonuclease to cleave the dye as if it were repairing an abasic lesion, resulting in a fluorescent signal whose color indicates the SNP call.
Though the assays outlined above represent some of the most popular genotyping options, they are not the only ones (see, for instance, Ragoussis' recent review). One emerging trend is using massively parallel DNA sequencing not just for SNP discovery (as in the 1000 Genomes Project), but also for genotyping of both "novel and known SNPs," says Karen Li, applications specialist at life technologies (which offers the Applied Biosystems SOLiD 3 Plus sequencing system).
"For most applications, next-generation sequencing provides more information than arrays," says Li. "With sequencing, scientists are not limited to testing for known SNPs. Genome-wide association studies and the 1000 Genomes Project suggest the presence of rare SNPs that can only be detected by sequencing."
With so many options at both the low and high ends of the SNP market, says Chanock, researchers in the hunt for genetic associations have tools available to meet every SNP density, save one. "The market is missing a really good, fast, reliable way to do 200 to 1,000 SNPs in a cost-effective manner," he says.
Whichever option(s) you choose, Ragoussis offers these three suggestions. First, rigorously quality control your DNA; he recommends Life Technologies' fluorescent Quant-iT™ PicoGreen® reagent."If the quality is right, then 50% of the work is done, more or less, because normally most assays on the market are robust," he says. "If one follows the instructions, it works."
Second, establish a solid informatics, databasing, and LIMS infrastructure to store, retrieve, analyze, and log data and experimental conditions. Finally, consider the logistics of running a large genotyping operation. For instance, how many freezers will you need to store all the samples and reagents?
"One has to think about these issues in advance, particularly because very often, these issues are underestimated," Ragoussis says. "So a good logistics approach is essential."
References
1Lou, H. et al., "Fine mapping and functional analysis of a common variant in MSMB on chromosome 10q11.2 associated with prostate cancer susceptibility," Proceedings of the National Academy of Sciences, Vol. 106 No. 19 7933-7938, May 12, 2009.
2 Ragoussis J., "Genotyping technologies for genetic research," Annual Reviews of Genomics and Human Genetics, 10:117-33, 2009.