NEW YORK (GenomeWeb) – Crop researchers are increasingly turning to next-generation sequencing, citing the declining cost of the technology, the higher complexity of plant genomes compared to animal genomes, and a host of other reasons.
At the same time, vendors like Affymetrix and Illumina, who sell both sequencing and microarray products to crop researchers, say that SNP chips are still very much in demand, especially at private breeding companies that use them to track markers from generation to generation.
Despite this, multiple interviews with some of the leading academics in the field of crop research revealed a preference for NGS, and genotyping-by-sequencing (GBS), in particular.
"I became interested in GBS approaches because of the high start-up costs of SNP arrays and the fact that maize, the species that I study in my academic life, is highly diverse," said Patrick Schnable, director of the Center for Plant Genomics at Iowa State University.
Schnable was among the early adopters of sequencing in the crop research community, moving to the technology more than half a decade ago after relying on high-density microarrays to study maize. Yet arrays, given their set, predetermined content, plus the expense of developing the markers for each chip, were unable to provide Schnable with the ability to unlock the diversity of various maize cultivars, many of which were genetically unique, meaning that an array developed for one line of maize would find limited utility in studying another line of maize.
"Of significant concern is that if the reference population differs from the target population, ascertainment bias will be introduced, as sites that were not polymorphic in the reference population won't be assayed even if they are polymorphic in the target population," commented Schnable. "This ascertainment bias can cause significant problems with the interpretation of genotyping data generated using SNP arrays," he said. GBS approaches, on the other hand, avoid this bias by "coupling SNP discovery with genotyping and doing both in the target population."
To help crop researchers make better use of genomic tools, Schnable in 2010 co-founded Data2Bio, a privately held company based in Ames, Iowa, that offers services such as SNP discovery, RNA-seq, high-throughput genotyping, gene mapping, and breeding support. Data2Bio also offers its own highly targeted GBS service called tunable genotyping-by-sequencing, or tGBS. According to Schnable, based on his experiences at Data2Bio, GBS is becoming increasingly favored by crop researchers worldwide, especially as the cost of sequencing declines and throughput increases.
"It is seldom cost effective to create arrays for crop species that lack the large research base of crops like maize and soybean," Schnable said. He noted Data2BIO has generated data for many less-researched crops, as well as for other research communities, such as ecologists, for whom designing a SNP array would not be cost effective.
Schnable acknowledged, though, that SNP chips remain the dominant technology for livestock research. He attributed this to the higher cost of livestock compared to plants, making it feasible to genotype cattle given the relatively low cost of a SNP chip versus the price of an animal. He also pointed out the relative lack of genetic diversity in livestock animals, making the development of SNP chips for various livestock species more attractive.
Another factor that has made SNP chips a mainstay of livestock genomics was the fact that livestock communities, cattle in particular, adopted SNP arrays years before GBS became available, meaning they are already heavily invested in the technology.
"SNP array technology benefitted from the first-to-market phenomenon," said Schnable. "Users had already collected substantial amounts of data using the SNP arrays," he said. "Many were unwilling to change technology platforms, though we are beginning to see signs of interest in GBS from the livestock community and I expect this to continue."
GBS is not the only technology that crop researchers are relying on, nor is their sole alternative SNP arrays, Schnable noted. He referred to lower-multiplex approaches marketed by Fluidigm, Sequenom, and LGC as genotyping options for crop research studies.
"Most of these approaches have a sweet spot with regard to the number of individuals to be genotyped and the number of markers to be assayed per individual," said Schnable. "But at this point for large numbers of markers GBS remains quite attractive."
"GBS is becoming much more popular," said Jackson, whose research is focused on peanut, soybean, common bean, cowpea, pigeon pea, and rice. "We use GBS for discovery that is winnowed down to informative markers for selection — both foreground and background selection," he said. As part of the selection process — during which researchers will use markers to make decisions about what lines to breed given their determined genetic characteristics — UGA researchers may rely on various other technologies, Jackson noted.
"It depends on the crop and the ones doing the breeding," said Jackson. "In the soybean program at UGA, they use Fluidigm and a smaller set of markers," he said. "They are exploring genomic selection with many more markers, but I don't think they have settled on either GBS or arrays at this point," he said. "It is clear that GBS keeps getting better and cheaper," he added.
That cost advantage may be spurring increased uptake of GBS in developing countries, noted Peter Wenzl of the Global Crop Diversity Trust, a Bonn, Germany-based organization that seeks to conserve and disseminate crop diversity for food security worldwide. Wenzl is the Crop Trust's liaison to DivSeek, an international effort that supports crop genomic research activities.
"In developing countries, NGS is leapfrogging older technologies, like with mobile phones," said Wenzl. As scientists in developing countries are just now beginning to invest in genomic technologies, they are opting for NGS, he noted, given its increased use in crop research. It's a trend Wenzl experienced firsthand as leader of Seeds of Discovery, an International Maize and Wheat Improvement Center (CIMMYT) project that relied on NGS to profile CIMMYT's maize and wheat seedbank holdings at its headquarters in Mexico City.
"I helped to set up a GBS facility in Mexico, there is a GBS facility in Kenya, and I hear talk of something similar in India," said Wenzl. "These scientists want to leapfrog to the next level rather than bothering with older technologies."
Wenzl noted that the use of technology depends on the application, and said that in the future, as crop researchers get a better handle on their species' genetic variation, they may very well move to SNP arrays. But for current seed bank accession, Wenzl said that GBS has the advantages of simultaneous discovery and calling of genetic polymorphisms. "It does it in a single step, making it inherently a better method to deal with a crop with large genetic diversity," said Wenzl.
Jesse Poland, an associate professor at Kansas State University, agreed with Wenzl that labs in developing countries are adopting GBS for their research needs. "If there was no prior investment in SNP arrays then it definitely makes sense to go directly to GBS," he said. But rather than using GBS as an upstream discovery tool, Poland said that he is using the approach directly in downstream genomic selection. "GBS markers work great for selection," Poland noted.
The challenge for Poland is not in technical issues related to GBS, but in implementing GBS-based selection directly into breeding programs. "Much of the challenge is logistics of everything moving very quickly in the breeding programs and not so much on the technical challenge of genotyping or sequencing," he said.
"With any new technology, there are many kinks to iron out," said Sarah Ayling, a group leader at The Centre for Applied Genomics in Norwich, UK. "Getting the confidence and familiarity with using the technology, and then using it in the applied field — that is something that will take time," she said. "I know that chips didn't happen overnight, either."
Still in its infancy?
While many academics have moved to GBS, companies that sell tools for crop research present a more nuanced view of the market, especially since they deal with the private crop breeding companies that, though they have adopted genomic tools en masse, seldom publish their findings. That means that they are largely absent from public discussions about technology use, according to Mike Thompson, director of global sales at San Diego-based Illumina.
"The large seed companies are doing an enormous amount of genomics work," said Mike Thompson. "It is not a secret but by and large they are not publicizing what they are doing with genomic tools," he said.
What they are doing with genomic tools appears to be routine work using SNP arrays. Thompson noted that Illumina has produced 17 arrays for crop research, each for a different species. "I would say that arrays are the predominant tool of choice for genotyping," said Thompson. "There is still huge usage of arrays," he continued. "People got standardized on the Infinium assay and it is turnkey for them to use it."
As for GBS, Thompson described the application as "still in its infancy," citing a few hindrances to widespread adoption, including data analysis challenges. "There is a lot of intellectual momentum," said Thompson, "but people are still getting hung up on the informatics."
However, Thompson also noted the trend in Asia toward embracing NGS, and GBS in particular, rather than investing in SNP array technology. "There is more momentum in Asia and India for using sequencing for everything and skipping the array step as they have no legacy systems," said Thompson. "That's one regional difference."
Looking ahead, Thompson predicted that crop breeders will continue the trend of using SNP array and GBS data for genomic selection. He added, though, that researchers will likely develop an interest in markers other than SNPs, such as insertion/deletion and methylation markers, for use in selection, which in turn could encourage more breeders to use GBS rather than arrays.
"That will drive momentum toward sequencing, and more and more crops will use predictive breeding instead of field trials," said Thompson.
Santa Clara, California-based Affymetrix is also aware of a growing market appetite for GBS. In May, the company acquired Eureka Genomics for $15 million in an effort to expand its menu for agricultural research. Eureka's technology enables the simultaneous profiling of SNPs and other genetic markers in hundreds to thousands of animal or plant samples using NGS.
With Eureka on board, Affymetrix is "building a beachhead in agrigenomics," said Laurent Bellon, senior vice president and general manager of Affymetrix's genotyping business unit..
According to Bellon, there are actually two GBS application areas: non-targeted and targeted. Non-targeted, which relies on restriction enzymes for genome complexity reduction, followed by sequencing, is considered to be a discovery tool, while targeted GBS, as offered by Eureka Genomics, allows users to interrogate very specific regions or mutations in the genome.
"What we see is conventional GBS used in academic labs on a smaller number of samples," said Bellon. "Once you get knowledge about traits, then arrays and targeted GBS are more mainstay," he said. "It is fair to say that conventional GBS is upstream to genomic selection, while more downstream applications are done using arrays and targeted GBS."
Like Thompson, Bellon said that crop breeding companies are using SNP arrays for much of their genotyping. "All of the major ag science companies are using arrays," said Bellon. "Arrays are being used as a main tool to follow markers from generation to generation and to select against that," he said, adding that arrays are "used routinely in marker-assisted selection programs."
One company that is arguably well-positioned to gauge the market's appetite for various technologies is Trait Genetics, a Gatersleben, Germany-based company focused on plant breeding and research. Founder and CEO Martin Ganal said that Trait Genetics currently offers Affymetrix and Illumina SNP array platforms in addition to NGS and Sanger sequencing for SNP identification. According to Ganal, GBS is attractive to crop researchers for a number of reasons.
"The main advantage of GBS is that it does not require a priori investments in a genotyping technology, no initial marker identification, no major investment in the development of an array," said Ganal. "Genetic maps can be created quickly and relatively cheaply."
He also noted that crop researchers favor GBS because they run fewer samples, in general, compared to livestock researchers, and because the genomes of their species of interest are typically far more complex. Wheat, for example, has a hexaploid genome, whereas all livestock genomes are diploid.
GBS is also a "great tool to identify markers in a less-investigated species," said Ganal. Still, he acknowledged that while academics favor the application, breeding companies have yet to adopt it.
"The main reason is that the associated bioinformatics is complicated," said Ganal. "This concerns especially the development of respective databases with marker information … when it needs to be maintained over a long period of time as during genomic selection, when data over many lines and years have to be collected," he said.
Additionally, Ganal said that in a well-investigated species, such as maize or wheat, the advantage of GBS is usually lost. "With the development of new optimized and haplotype-specific arrays, the costs of array-based marker analyses have come down to the same level as GBS, and less redundant information is being generated," he said. Ganal said that some of Trait Genetics' customers have even returned to using arrays instead of GBS because of the high reproducibility of the data generated.
There are also IP issues related to using and offering GBS, Ganal acknowledged, another issue that might slow adoption of the approach by breeding companies.
"Many if not all of the GBS technologies are covered by one or several patents and one might need several licenses, one for research and one for commercial use, for example, to [use] the technology," Ganal said.
While there may be impediments to GBS adoption among breeding companies, Allen Nguyen, associate director of business development for genetic sciences at Thermo Fisher Scientific, does believe a movement toward GBS is underway across the agbio market.
"GBS has become a very hot technology in the plant and animal research and breeding markets in recent years as demonstrated by the growing number of talks and posters at recent Plant and Animal Genome conferences," Nguyen said.
"Because it is a newer technology than microarrays or quantitative PCR, GBS using NGS platforms may not be as prevalent or widely used as those older tools, but we see a technology shift taking place in the market," Nguyen continued. "Several people at very large agbio companies have indicated that our Ion Ampliseq GBS technology could really transform the way they do their genomic selection and screening," he added, referring to the company's kits that run on its Ion Proton and Ion PGM sequencing instruments.
Like Affymetrix's Bellon, Nguyen positioned the firm's Ion Ampliseq GBS kits for downstream genomic selection and screening. "Where researchers are narrowing their marker sets and targeting known SNPs, it is optimal for analyzing hundreds to thousands of targets per sample in a very efficient and robust way," Nguyen said. "Moreover, additional novel SNPs that are discovered in the same amplicon may be useful for breeding purposes."
Interest in Ion Ampliseq has not been limited to crop researchers, who tend to favor GBS compared to livestock researchers, Nguyen noted.
"In both the plant and animal markets we've seen a very fast growing interest in GBS and believe it has a very bright future to become a foundational technology for routine plant and animal breeding," he said.
Though GBS may eventually become the go-to application for genomic selection, companies that have traditionally positioned themselves downstream of sequencing and SNP arrays are ever ready to serve a market that is interested in surveying a set number of markers across many samples.
"The challenge of these GBS projects is that they have typically been with very small numbers of samples and they generate extremely large data sets," said Fluidigm spokesperson Howard High. "It's these type of projects that have enabled development of high-density chips in the past, but with NGS costs becoming even lower and many labs wanting to look at even more targeted regions, we are seeing Fluidigm's technology implemented in more NGS workflows," he said.
According to High, the company's Juno platform and its new Orion chemistry fill this emerging gap. The South San Francisco, California-based firm launched Juno earlier this year. The system allows researchers to genotype input DNA and get 9,216 data points in about three hours.
"With all the data being generated … researchers can find it difficult to separate the wheat from the chaff," said High. "They need to cut down on the total number they need to use in order to further reduce their costs," he said. "This seems to hold true across all non-human species — crops, aquaculture, and livestock."