Studies have found genes associated with a significantly higher risk of severe COVID-19, some of them claiming it to be 1.6 times the general risk.

They mention genes at markers such as rs35044562, rs1129183, rs679599192 to be significant.

Can people use the raw data from a consumer-grade genetic test like 23andMe to assess their risk, at least theoretically? The raw data has many gaps and the markers and their positions may not be included in the data. Can one then look at markers at positions close by, or at other markers? Does a more rare genotype imply higher risk?


  • 5
    $\begingroup$ I would strongly recommend talking to a genetic counselor about any results found on a genetic test, whether it be about COVID-19 susceptibility or anything else. Also, please keep in mind that personal medical questions and health advice are off-topic on Biology.SE. We cannot safely answer questions for your specific situation and you should always consult a doctor for medical advice. This question is generic, so I'm not going to vote to close it, but I just wanted to make you aware in case you were planning on doing this yourself. $\endgroup$
    – MattDMo
    Dec 8, 2020 at 19:03

1 Answer 1


I suppose the answer is 'theoretically yes, but don't trust the results much'.

The 23andme genotyping array is very useful for ancestry testing, but much less useful for clinical diagnostics. I have worked in the past for Genomics England (an arm of the NHS in the U.K. which performs genetic diagnostics for rare disease). There, the samples were sequenced to very high coverage and underwent extremely stringent quality control checks which were being constantly updated. If you want to interpret a variant in a clinical sense, you have to very sure that you get the right genotype.

I believe that the details of how 23andme performs its variant calling and quality control are proprietary and therefore I would never trust their variant calls for any kind of clinical diagnostics, and neither would any clinician I know. This doesn't mean 23andme results are invalid - they are very useful and pretty accurate for testing ancestry, but they are not valid for clinical diagnostics. If your risk calculations hinges upon a few SNPs of relatively large effect size (such as BRCA), then you want to be very sure you make the correct variant call. You simply do not get this assurance with 23andme.

Secondly, like you mention, many of the SNPs you refer to aren't covered by the 23andme array. The effect size of a particular SNP on a trait is calculated for that specific SNP only. Extrapolating the effect size for one SNP in a gene to another, nearby, SNP in the same gene is likely to give you innaccurate results. In fact, the same SNP in the same gene can have pretty different effects in different people (https://febs.onlinelibrary.wiley.com/doi/pdf/10.1111/febs.14080 caused by the environment, pleiotropy).

  • $\begingroup$ Curious as to why this was downvoted? $\endgroup$
    – user438383
    Dec 9, 2020 at 13:12
  • $\begingroup$ No idea. It's exactly the sort of answer this type of question needs. $\endgroup$
    – MattDMo
    Dec 10, 2020 at 13:56

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .