I recently started analyzing a GWAS summary dataset. It has a column for the reference allele in which it mentions the base in the reference genome.

There is another column for the alternate allele (or the SNP). Each reference allele always has one alternate instead of three. I have read responses to this question on other forums and the answer usually goes like this:

The other two alternate alleles are very rare so not seen in the population that often.

I find this hard to believe.

On average every person has 3,000,000 SNPs (1 in 1000 bps). Does that mean in all the 3 million cases the individual always has that one alternate allele? Let's take a million individuals, do all those 3 million million SNPs always have that one alternate allele?

This just seems very unlikely, even if those alleles are very rare we have so many occurrences that we should be seeing them.

I am new to GWAS and have taken genetics courses during my undergrad. Could someone provide a satisfactory explanation? Sometimes I feel we are doing this as an approximation to make further statistical/computational analyses easier.

Edit: Although I am asking this in the context of GWAS, I have seen it in many other situations. It is sometimes referred to as the minor allele.


1 Answer 1


Estimates seem to be around 45-60 de novo mutations in an individual, e.g.: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC9550265/

That means the other 2999950 mutations are inherited, they're shared with a parent and their parent and possibly siblings, cousins, so on. They won't have inherited just some random base pair at that location, they will have inherited exactly the same one as all those other people share with them.

Additionally, there are some SNPs that are going to be selected against. Any SNPs that are lethal will never be inherited, for example.

So, overall, I don't find it that surprising. Less common SNPs that haven't spread in the population are unlikely to be found in any sample of thousands or hundreds of thousands of humans in a population of billions. So, they may be out there but never observed. And unless they are observed they aren't very useful for the things people want to use SNPs for.


You must log in to answer this question.

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