High-density SNP arrays are coming, and Silicon Genetics is ready. This week, the company will launch what appears to be the first commercial software package for whole-genome genotyping experiments.
Silicon Genetics’ CEO Saeid Akhtari said that the software, called Varia, has been under development for nearly a year and a half. The company has already proven that first movers actually have an advantage in the bioinformatics market — GeneSpring, the gene expression analysis software that it launched in 1998, remains a perennial favorite, with over 4,000 users worldwide, despite an explosion of competing products and companies since its launch. Silicon Genetics is counting on a similar effect this time around, and predicts that the market potential for genotyping technology — and, subsequently, genotyping software — could outstrip that of gene expression.
“I personally believe the genotyping market is where the gene expression market was in 1998, except I don’t believe it’s going to take as many years to get to where the gene expression market is today, in terms of size and usage,” Akhtari said. Many of the “technical issues” that plagued microarray technology in the early days have since been worked out; the price of genotyping microarrays is much lower than that of gene expression chips several years ago; and many labs have already invested in the infrastructure to run expression arrays, he pointed out, so the barrier to entry for running genotyping chips is far lower. Finally, Akhtari said, the large genetics community is a potential new user base for genotyping tools. “There are so many scientists interested in the study of genetic variation in humans at a single-base resolution,” he said. “I believe there are more people interested in that than gene expression.”
Greg Yap, senior director of DNA analysis at Affymetrix, said that the market for its genotyping platform “has started to pick up.” While he did not provide specifics on what kind of growth the company anticipates for the genotyping market, he said that it should be “roughly the size of the gene expression market, and it’s growing quickly.”
As it did with GeneSpring, Silicon Genetics has in this instance bundled a number of commonly used algorithms and statistical techniques into a single package designed especially for very large data sets. The software includes algorithms for linkage analysis, pedigree analysis, haplotype mapping, and data visualization. Kevin Wandryk, vice president of marketing and business development for the company, noted that “the basic understanding of this kind of genotyping analysis isn’t new — after all, that’s what Mendel discovered years ago with the variation in peas.” What is new, however, is the amount of genetic variation data produced by a single experiment. Affymetrix’s Mapping 10K Array, launched last July, can analyze 10,000 SNPs at a time, and the company has already began shipping its 100K array, which analyzes 100,000 SNPs, to early access customers.
“A lot of the earlier public domain software just wasn’t geared to handle tens of thousands of data points across hundreds of samples at one time,” said Wandryk. “So we’ve been really cognizant of building from the ground up a tool that’s capable of handling millions of data points on a desktop computer.”
Beta testers for Varia told BioInform that the software appears to live up to the company’s promise — with a few caveats. Frank Middleton, an assistant professor in the department of neuroscience and physiology and director of the microarray core facility at the State University of New York Upstate Medical University, said that colleagues in a population-based study spent months manually tracking down the location of a particular mutation, because publicly available tools like GeneHunter and Merlin weren’t able to analyze the data. In the meantime, he said, the lab installed Varia, loaded in the data, and found the same region almost immediately. The group has just submitted a paper on its findings.
Middleton said the software could be “an experimental biologist’s dream and a statistical geneticist’s nightmare,” because it may eliminate the need to “pay a statistician to analyze your data through all these different algorithms.” The only capability that Varia appeared to be lacking, he said, was for non-parametric analyses of data, in which the user does not provide any assumptions about patterns of inheritance. For that, Middleton said, he will probably continue using Merlin.
Deitrich Stephan, senior investigator and director of the neurogenomics division of the Translational Genomics Research Institute, said that Varia is the “first software package I’ve seen that is a unified solution” for genotyping analysis. Stephan’s lab was previously using publicly available software developed for low-density microsatellite markers; but, he noted, when the lab’s experiments ramped up from 400 markers to 11,500, “a lot of software programs we used bogged down.” While pleased with Varia’s performance, Stephan noted that it’s “fairly pricey” for his lab’s budget.
Silicon Genetics did not provide pricing information for Varia, but Wandryk said that because the market for genotyping software is still “pretty narrow,” it is priced higher than GeneSpring, which currently costs $8,000 per copy per year for a commercial license.
Competition at the Gates
Silicon Genetics does seem to have a jump on its potential competitors in the genotyping software market, but it’s likely that the company won’t have as much of a head start as it did for GeneSpring. At least one of the company’s competitors in the gene expression space is already working on a SNP analysis package. Jason Goncalves, Iobion’s CSO, told BioInform that his firm is “actively developing” a genotyping software package that will integrate with the Affy platform as well as with Iobion’s other products. Goncalves did not provide an expected launch date for the software.
Silicon Genetics will face competition from another front as well — the academic community, where developers are already modifying freeware for gene expression analysis to handle genotyping data. Wing Wong, professor of computational biology and biostatistics at Harvard School of Public Health and developer of the publicly available dChip gene expression software package, said that he and his colleagues are currently working on a new module called dChipSNP for loss-of-heterozygosity studies using SNP array data. A paper describing the module is in press at Bioinformatics. Wong said the DChipSNP module should be available online “in a couple of months.”
Right Place, Right Time?
The Affy 10K chip — along with a number of other high-throughput genotyping platforms — has been on the market for some time now, but Affy’s Yap said that it’s not unusual for commercial software packages to lag a bit behind new technologies. “Initially, when any technology comes out that provides new levels of data for analysis, the first sets of tools are from nimble academic groups … and the commercial companies come in behind that and put more of a user-friendly package around some of the algorithms,” he said.
Wong said that he wasn’t familiar with Varia, but noted that “there is a very large need there for software that can make it easy for people to use the Affymetrix [SNP] arrays. … Basically all the components are there in the field, but nobody has really put it all together into software that is easy to use and powerful and flexible.”
Wong added that the research community has been slow to embrace SNP arrays so far, because “the density has not yet reached the critical number to make it worthwhile. But now we are seeing that with the new technology, there’s a chance that it will be very widely used, so that basically motivates groups like us to try and do something about it.”
If genotyping does take hold in the genomics and genetics research communities, it’s likely that there will be plenty of room in the market for Silicon Genetics as well as its competitors. In addition, Akhtari noted, there’s room for growth left in the gene expression analysis market, which is still “relatively new.” The number of GeneSpring citations in the scientific literature rose from 1 in 1999 to 59 in 2002, and then shot up to 228 in 2003, he said. Currently, “We’re getting around one per day, so we expect the technology to continue to grow at a very rapid rate for several years to come,” he said.