It’s no secret that DNA microarrays have been overshadowed by next-generation sequencing (NGS). Many people beieve they have had their heyday, and now it’s time to embrace modernity and make way for their technological successors. Yet there are still people who prefer things simpler and there are even instances when it makes sense not to abandon the tried and true for the fast and fancy.
The basic concept of a microarray is a solid support to which oligonucleotide probes are attached to create an ordered array. Labeled samples are allowed to hybridize with the probes, the array is imaged, and the location of the label indicates the presence (or quantity) of complementary sequences in the sample. Expression arrays will typically query cDNA (RNA’s proxy), often comparing samples taken across different times or treatments, for example. Genotyping arrays, on the other hand, look in genomic DNA for the presence of single nucleotide polymorphisms (SNPs), for example, or structural abnormalities such as inversions, deletions, and duplications.
Here we look at some older, and some modern, applications in which microarrays prove their worth.
Express yourself
“For gene expression,” Steven Head, director of the NGS and Microarray Core Facility at the Scripps Research Institute “can’t think of anything you can do with arrays that you can’t do with RNA-seq”—a view shared by many core directors consulted for this article.
Reviewers of top-tier journals increasingly expect RNA-seq data for expression analysis, notes Stephen Baylin, the Johns Hopkins Medical Institute Ludwig Professor of Cancer Biology.
But that’s not to say that there’s nothing to recommend the microarray.
“Arrays still have significant benefits over RNA-seq,” points out Heidi Kijenski, director of clinical marketing, Agilent Technologies. For example, the workflow is much simpler, which reduces the technical expertise required. Microarrays can yield results quicker—a boon for CROs and clinical labs. Arrays boast a lower cost per sample, especially when running larger studies. And the data analysis (and data storage) requirements are significantly lower—it’s not necessary to have a bioinformatician on the team to interpret the results.
There are still many customers doing microarrays, according to a major university core director. Affymetrix offers a “very good, updated gene chip called Clariom D. Because of the ease with which it is done, customers can get data very quickly, and analysis of it is pretty straight forward.”
NGS and microarrays may also be used in conjunction. For example, “customization capabilities of the Agilent Gene Expression platform enable researchers to follow-up RNA-seq experiments, including any new transcript of interest, and balances experimental time and costs without compromising performance,” notes Valentina Maran, Agilent’s global product manager of cytogenetics.
On the flip side, Baylin will often run preliminary experiments on microarrays to get the lay of the land, looking for genome-wide expression changes before proceeding on to NGS: “It gives us a feeling for the directions in the gene changes and pathways we may be dealing with.”
It's in (or on) the DNA
Microarrays still play a significant role in clinical testing. Biomarker panels to genotype cancer or for transplantation tissue, for example, “took years to validate and get running and get approved,” points out Head.
Another big genotyping application is genome-wide association studies (GWAS), but even those are getting supplanted by NGS. Head explains that large databases of heavily sequenced populations allowing the use of shallower sequencing, combined with tools for calling and imputing genotypes based on haplotypes, has brought the cost of sequencing down significantly.
He notes that niche applications of DNA microarrays—such as those interrogating the microbiome, or CpG islands, or even DNA-protein interaction—can also be found.
Yet another specialized application seems to be breaking out from the pack. “The big perpetuation of microarray type of work in our work is for the genome-wide patterns of DNA methylation, using the EPIC arrays from Illumina,” says Baylin. 80–90% of utilization and billing at the DNA Microarray Core Facility, of which Baylin was until recently the faculty director, is from methylation arrays. “It’s a real staple for us … and I think many, many others have found the same.” While bisulfite sequencing may yield similar information, complex analysis and informatics make it more costly.
The currently available EPIC array (the Infinium MethylationEPIC BeadChip Kit ) is designed to query human samples, but “the mouse one is supposedly coming along,” says Baylin. “I think that’s going to get tremendous use, too. Because right now we do a lot of correlative mouse studies, but up to now we couldn’t use arrays.”
Infrastructure spiral
In some ways, the DNA microarray is caught in a vicious spiral. The advent of alternative technologies like NGS has made the microarray less essential to probing the transcriptome and genome. With the demand for sequencing services rising and the demand for microarray processing and analysis decreasing, many institutions are combining their core facilities or even eliminating microarray cores altogether. As this happens, a core director said, in order to support their product, manufacturers such as Affymetrix are now offering CRO-like processing services, further decreasing the demand for those services by institutions, causing more to consolidate or close. And as local support becomes harder to come by, time-to-result tends to increase, further driving away researchers.
Maintaining the infrastructure necessary to support microarrays is costly in terms of space, equipment upkeep, and personnel training. Some cores have succumbed, others have persevered (“We are not as busy as before, but still busy enough not to close down,” said one core director). Baylin’s core is currently weighing the benefits of local quality control and turnaround time against those of outsourcing. Either way, though, he will be using the methylation array “until something better comes along.”