I've heard about services like 23andme, which offer genetic testing to the general public. As a person who knows very little about genetics, I'm interested in the subject and would like to know what the modern "commercial grade" genotyping really does. What are the limitations of commercial-grade "genotyping"? From what I've been reading, they test DNA against ~100 different markers, but do not really sequence it.

If I understand correctly, if DNA is sequenced, genetic markers and genes can be identified in it, including the newly discovered ones. However, (if I understand correctly) if DNA is genotyped, the genotyping is a one shot operation which states if some genetics markers are present in the DNA. To test against newly discovered markers, it would have to be genotyped again, right?

Thank you for any clarifications!


3 Answers 3


23andme briefly describes the technology they use here. They are testing the genotype of your DNA at roughly 1 million locations. The technology they use to do this is known as a microarray.

The limitations of using a microarray, as compared to sequencing, is that you will only find what you are looking for -- people often describe the disadvantages of microarrays as compared to sequencing with the streetlight effect/metaphor.

Arrays can only measure regions of the genome that they were designed to probe. Technically, if they probe 1,000,000 million locations, and the human genome is roughly 3.4 billion bases ... you can do the math.

In practice, SNPs tell you a bit more about just the nucleotide being interrogated, due to linkage disequilibrium (cf. tag/proxy SNPs), so the array might tell you more than you expect.

Of course, modulo sequencing errors, whole genome sequencing will tell you "everything" as far as recovering the NTs that make up your genome, but how much information that will provide you is another thing altogether (for now, that is).

  • $\begingroup$ +1: on the other hand, 23andme proposes a certain type of analysis to their clients that does not require whole genome sequencing. They say "We will test such and such SNP and tell you that you have x% probability of having an heart attack in your 60s"... $\endgroup$
    – nico
    Feb 21, 2013 at 8:09
  • $\begingroup$ @nico: Yeah, very true. You can also get a copy of the data, though, which is cool. Also, they're also currently offering exome capture for some already existing members ... a look towards what the future may hold, perhaps. $\endgroup$ Feb 21, 2013 at 15:23

Just to be transparent, I work at a microarray company, but I've also attended high throughput sequencing workshops. I think this is a pretty fair assessment, but leave me a comment if you see anything out of line...

As pointed out by @SteveLianoglou, Genotyping microarrays are best for revealing whether a particular variant (not only SNPs but short insertions and deletions, copy number variations (CNVs) reliably and cheaply.

High Throughput DNA Sequencing has made great strides and the latest cost estimates are in the few thousands of dollars for a full human genome. This is better for discovering new variants but currently its not a great way to do do say a medical test to see if you have a particular mutation. The problem with sequencing right now is that sequencers have systematic biases - the local sequence can give the same error over and over again. It takes some significant effort to differentiate the error from the real variant data. In the workshop I attended, the data consisted of 38 full human genomes at 30-40x coverage. Despite this deep level of sequencing, systematic errors were still obvious (the proceedings are in press now - reference should show up at the link provided soon). So it can be fairly expensive when you have done all your data analysis and validation data to get the variants via sequencing. All this could change very quickly, but I think this is accurate in the current state of things.


I want to give the OP a slight variation on Steve's answer.

It is important to note that most of the genome is the same between different people. That isn't surprising, since we belong to the same species.

For example, it is estimated that there are about 10,000,000 SNPs (single basepair positions that may vary) in the human genome. So if you check 1,000,000 SNPs with a microarray (genotyping), you are covering ~10% of the possible differences. In addition, variation in some of the SNPs is very common while others are very rare, so if you check the most common 1,000,000 SNPs with a microarray, you might actually cover a very large portion of the differences. It is also possible that if you know in advance some general information about the person you are testing, such as ethnicity, you might be able to eliminate many uninformative SNPs.

This of course only refers to variation in SNPs (but these are the most common variations).

  • $\begingroup$ This is a good point - to test only what varies the most $\endgroup$
    – Alex Stone
    Feb 21, 2013 at 14:56

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.