Less than a month after Illumina launched its GoldenGate methylation product and accompanying 1,500-CpG loci cancer panel, CEO and President Jay Flatley hinted that the firm is adding products to its methylation portfolio and could release them soon.
The company is not alone.
So far, most methylation arrays on the market focus on CpG islands. But it is beginning to look like that situation will change in the coming year as customers begin demanding chips that have more useful content, are more flexible, and come with methylation-specific bioinformatics tools to help analyze data.
Illumina, for its part, in early January debuted a high-throughput DNA methylation-profiling technology that can survey up to 1,500 sites across 96 samples simultaneously. The first standard product in this portfolio is a cancer panel covering 1,500 sites over 800 cancer genes. “Custom content panels will soon be available to meet individual researcher needs,” Flatley said during Illumina’s fourth-quarter conference call this month.
In a follow-up e-mail to BioArray News, Flatley wrote that Illumina will release additional standard panels, like the cancer panel. He declined to elaborate.
This month both NimbleGen Systems and Agilent Technologies also said that they have new methylation designs in R&D. Both companies declined to elaborate.
Cater to the Customer
Even in a market as nascent as the one supporting methylation arrays, customers are already starting to use different methylation tools that fit their specific research goals.
For instance, Peter Laird, a molecular biologist at the University of Southern California and an Illumina customer, focuses mostly on DNA-methylation changes that occur in cancer. “The current commercial panel from Illumina has many, but not all, known DNA-methylation markers of interest,” Laird told BioArray News in an e-mail last week.
“The flexibility and customizability of the Illumina GoldenGate DNA methylation platform will allow evolution and specialization of oligo sets in the next few years,” he wrote. “So non-informative markers currently on the array will be omitted, replaced by … informative markers.”
Laird added that he believes “specialized oligo sets will emerge for different types of cancer, or that address specific biological questions such as genomic imprinting.”
Over the long term, Laird wrote, he sees Illumina’s “product mix evolving into a collection of generic panels and specialized panels, each of which have been optimized to exclude non-informative markers.”
Whole-Genome vs. Focused Arrays
Like Laird, users of platforms from other vendors see new products as key to the expansion of methylation arrays — but in different ways.
Hiroyuki Aburatani, a professor at the Research Center for Advanced Science and Technology at the University of Tokyo and an Affymetrix user, told BioArray News last week that he believes higher-density arrays, as well as arrays with more content, will be will play a critical role in transforming array-based methylation studies.
“Because of the limitation of probe numbers, [other companies like] Agilent and NimbleGen mainly focused on CpG islands, which would be functionally important in transcriptional regulation and frequently found to be aberrantly methylated,” Aburatani wrote.
In Aburatani’s opinion the promoter regions “should be well covered” in methylation analysis. “We still do not have enough information on the exact locations of promoters for all the genes, particularly because each gene often may have more than promoters,” he wrote. “To elucidate the role of methylation in transcriptional regulation, you need to analyze them together.”
“Current information on the promoter location is still limited, [but] a combination of promoter and CpG island sequences will be nice to have on a single array,” he wrote.
Aburatani also said that Affy needs to develop higher-density and more-affordable arrays in order to provide researchers with more power. “Right now, whole-genome tiling arrays can cover both promoters and CpG islands. However, unless much higher-density arrays become available, current whole-genome tiling arrays are not suitable to analyze many samples,” he wrote.
In its most recent earnings call, Affy said that methylation services were currently being offered through service providers such as Genpathway.
Mary Harper Warren, chief scientific officer at Genpathway, which recently launched a methylation-profiling service based on its whole-genome tiling platform, agreed with Aburatani that density is certainly an issue that companies in the space need to address.
“To be truly useful, a methylation assay must achieve whole-genome coverage,” she wrote in an e-mail to BioArray News this week. “Any array and/or assay that surveys only CpG islands, promoters, or the relatively small percentage of the genome currently known to be methylated, is limited.
“It is not feasible to ‘add’ specific content to methylation arrays [or] assays and make them more useful without going whole-genome,” Warren wrote.
Achim Plum, vice president of corporate communications at Berlin-based Epigenomics, told BioArray News this week that his firm, which offers its differential methylation hybridization arrays to partners as a service, agrees that researchers need broader tools.
“We believe that looking at every single CpG in the genome will not help to extract relevant data,” he wrote. “Broad DNA methylation approaches ... will help to better characterize which regions of the genome have the potential to carry functionally relevant DNA methylation information.”
”On the other hand, the results of DNA methylation research lead to diagnostic applications of specific marker panels,” according to Plum. “Here, customized microarrays will help to select the best marker in research and finally will be used as US Food and Drug Administration-approved in vitro diagnostics.”
“Broad DNA methylation approaches ... will help to better characterize which regions of the genome have the potential to carry functionally relevant DNA methylation information.”
But if screening arrays made by Affy or Epigenomics supposedly offer the widest coverage, then why are other researchers using different platforms? Dirk Schübeler, a group leader at the Friedrich Miescher Institute for Biomedical Research in Basel, Switzerland, said he uses NimbleGen arrays because the company customizes arrays for its clients.
“We are mainly interested in looking at promoter DNA methylation in the human system, also in the mouse. So far these have been extremely accurate arrays,” he told BioArray News this week. Schübeler added that he is aware of the whole-genome offering from Affymetrix, but that the cost of using the arrays is prohibitive.
“Right now it’s possible to do the whole genome, but that would mean 10 to 12 arrays per experiment,” he said. “That makes it unfeasible.”
NimbleGen also plans to launch whole-genome tiling arrays, and expects to debut its first global methylation-profiling method “within the next month,” Xinmin Zhang, the ChIP-chip and methylation business unit manager at NimbleGen, told BioArray News recently in an e-mail.
According to Zhang, the product will be “coupled with whole-genome sets of tiling arrays for human and most model organisms and any other organism available upon request, as well as a new suite of single array designs that include dense coverage of all known promoters and CpG islands in the genome.”
Some customers pine for more focused methylation arrays because of the challenge of analyzing the data produced by the new chips.
According to Schübeler, “once you start using a certain platform, then you invest quite a bit of getting bioinformatics set up that’s tailored towards these data sets. Right now a lot of these analysis tools are homemade. There aren’t many good software packages that you can just use with these arrays like you can with gene expression,” he added.
Michael Rehli, a scientist at Regensburg University Hospital in Germany who is also involved in array-based methylation research, agreed with Schübeler, saying, “I am definitely in need of some kind of bioinformatics tool to normalize the data. I honestly don’t care if it comes from the companies or a software provider.”
Rehli said that his lab has been trying to adapt existing programs for its analysis, but they “are all not really optimized for methylation. Most of the time we just end up correlating our own data,” Rehli told BioArray News this week.
Still, Rehli said he was confident that tools for methylation analysis will become available soon. ”This is a new market and it won’t take long for people to realize that there is a market — that epigenetics is very important in diagnostics, for example,” he said. “These tools are not very old. I guess it will take time to design [bioinformatics] tools that are used specifically for methylation analysis.”