Forum:Future of Microarrays

Question

[E]

Hello,

I have a more general question about microarrays.

Where do you see the microarray technology going? As far as I know it has basically been replaced by sequencing technologies when it comes to research. Do you know if there are still applications where microarrays are used or where microarrays outperform sequencing?

And how do you assess the value of existing data? Is it still used and if so in which context?

All the best, E

[Ventrilocus]

Whenever a large number (e.g. hundreds/thousands/ten thousands) of samples are used, NGS becomes just too expensive. For that reason, SNP arrays are the norm for genome-wide association studies (GWAS), DNA methylation microarrays are the norm for epigenome-wide association studies (EWAS) and RNA expression microarrays are the norm for transcriptome-wide association studies (TWAS). Also, they are highly used for meQTL and eQTL mapping. You'd be surprised how common they still are in spite of NGS overcoming almost all of its limitations.

[cpad0112]

Due to increase in commercial DTC-GTs, microarray market is still strong. For eg 23nme, daytwo, viome etc.

Answer 1

[GenoMax]

Arrays remain alive. Demise of arrays has been predicted for a few years but they continue to be used since they are a proven technology. Data analysis methods are robust and easily accessible via R packages or commercial software.

Is it still used and if so in which context?

Best example of continued use of arrays is probably in agrigenomics/animal breeding where they are used for screening genetic traits.

If you check on NCBI GEO (LINK) you will find plenty of recent data submissions. Under Find platforms at top, search using Manufacturer --> Affymetrix as an example.

Answer 2

[lieven.sterck]

in most/many aspects indeed NGS has replaced the use of microarrays, however there are still a few exceptions.

For well established species/genomes (eg. human, mouse, ...) that have well designed arrays they are likely still in use and might even be still more cost-effective than doing it via sequencing (but these are exceptions!) For most species , and especially the 'newer' ones they will not invest anymore in creating microarray chips and will thus solely rely on NGS approaches.

Then of course you also have things like SNParrays which I know are still quite frequently used but also here is the notice of the use-case I mentioned above in effect.

Answer 3

[Joe]

Outside of genomics, "microarrays" are also still used quite a lot in protein-protein and protein-ligand interactions. Glycan arrays are common for protein binding studies for example.

Since they work in broadly the same way, a lot of the same data analysis packages/approaches can still be used.

Answer 4

[4galaxy77]

IMO I hope that people stop using genotyping arrays in the near future given the evidence suggests you get a fair bit more power in e.g. GWAS or qtl mapping if you spend the same amount of money on low-coverage sequencing + imputation compared to genotyping arrays. This is especially the case for underpresented groups who don't have variants present on the most commonly used arrays & for finding rare variants. Low coverage sequencing also means it's much easier to co-analyse datasets which might otherwisw have been sequenced on different chips.

https://www.nature.com/articles/s41588-020-00756-0

https://genome.cshlp.org/content/31/4/529.short

Answer 5

[Kevin Blighe]

To quote the Affymetrix CEO, Frank R Witney, in the meeting that we had a number of years ago: 'We still believe in the microarray'.

Kevin

[Joe]

Yeah but he would say that ;)

Answer 6

[Papyrus]

To add another example: epigenetics. Just this year a new DNA methylation array has been developed to profile conserved mammalian CpGs (see bioRxiv).

A problem with NGS technologies which appears when profiling many samples, and is aggravated when comparing species (because one also has to lift genomic coordinates), is that, because they do not profile the "exact" same genomic locations, it becomes harder and harder to find common measured locations across all samples (i.e. the data is "sparse").

Arrays help provide a robust set of well-known locations in which to explore your study system. Yes, this also "limits" the scope of possible findings, but of course this will depend on the goal of the experiments. In this case, currently DNA methylation arrays are often used to profile epigenetic clocks and epigenetic age acceleration, and I suppose that this was the main motivation in developing the mammalian array (i.e. to be able to profile the same biomarkers across species).