I'm trying to get a sense of the dominant way that mutations occur. I have seen various numbers which seem at least at first glance to conflict, and I was curious if anyone had clarification on this.

According to Shastry, "SNP alleles in human disease and evolution" (Journal of Human Genetics 47:561–566, 2002),

In two randomly selected human genomes, 99.9% of the DNA sequence is identical. The remaining 0.1% of DNA contains sequence variations. The most common type of such variation is called a single-nucleotide polymorphism, or SNP.

However, according to Nevo, "Genetic Diversity" (Encyclopedia of Biodiversity, 2001),

Significantly, SSRs experience mutation at notably higher rates than do nonrepetitive sequences: $10^{-2}$ to $10^{-3}$ per locus, per gamete, per generation, which leads to their high polymorphism.

Additionally, according to Kashi and King, "Simple Sequence Repeats as Advantageous Mutators in Evolution", (TRENDS in Genetics 22(5):257),

Mutations that alter repeat number typically occur at rates orders of magnitude greater than single-nucleotide point mutations.

So, if variations in organisms is mostly due to SSRs, and this is mostly copy number variation, then would it be correct to say that copy number variants in SSRs is the dominant form of variation, or am I missing something? Which one is correct, or which source is more authoritative, or is there a more up-to-date review I should be consulting?

  • $\begingroup$ NOTE - I added the word "mutations" to the question title to be more clear what I'm asking about. $\endgroup$
    – johnnyb
    May 16, 2021 at 23:58

2 Answers 2


I like @mgkrebbs answer, I think that it hits most of the high points, but I wrote a review on this subject a couple of years ago where we specifically put together estimates on the mutational load of different mutation classes (see Table 1).

Yaniv Erlich's group directly addressed the question that you are trying to answer, and they estimated that microsatellites contribute slightly more mutations per generation than substitutions:

These predictions indicate that the load of de novo STR [SSR] mutations is at least 75 mutations per generation, rivaling the load of all other known variant types.

Most estimates of diploid overall substitution mutation rates are ~50/generation, for comparison.

However, if you are interested in overall variant number you will also need to take into account an incredibly complex architecture of structural variants in basically any genome, including also plasmids, viruses, and transposons. So when you talk about the "dominant form of variation", you really do need to be a bit more specific. On a per locus or even per mutation class basis, STRs/SSRs probably are "dominant", but it's hard to argue that e.g. 1 STR mutation is more important than one transposon hop or one chromosome aneuploidy or one centromeric satellite contraction.

I'd recommend reading the papers linked for a little more context.


It's not really clear to me what you're after when you ask about how prominent or dominant these two kinds of variations are. But let's look deeper.

Guichoux et al. (1) say:

There are two main differences between SSRs and SNPs. First, SNPs are more numerous than SSRs in the genome of most species. On average, in the human genome, there is one SNP every 100–300 bp (Thorisson et al. 2005), com-pared to one SSR locus every 2–30 kb (Webster et al. 2002), depending on how SSRs are defined (Kelkar et al. 2010).This can be important for genome-wide association studies but not necessarily for other applications. Second, the mutation rate per generation differs drastically between the two marker types. SSRs have mutation rates ranging from $10^3$ to $10^4$ per locus per generation (Ellegren 2000), compared to about $10^9$ for SNPs, i.e. several orders of magnitude lower. As a consequence, SNPs are typically diallelic: In humans, < 0.1% of SNPs are triallelic (Lai 2001).In contrast, SSR loci generally have high allelic richness, often in excess of 10 alleles.

Based on these numbers, SNPs occur perhaps 80 times as often as SSRs. Is that more prominent?

An SSR is larger than an SNP, covering perhaps one hundred to several hundred bases versus one base (by definition). An SSR locus also occurs in multiple variants, e.g. having a copy number from a wide range, versus only two possible variants that usually occur for an SNP locus.

The mutation rates are substantially different, apparently typically 5 orders of magnitude more frequent for SSRs than for SNPs. Note, though, that mutation rate does not directly affect the frequency of variants seen in the genomes of extant organisms: the variants present are a result of the filtering of mutations by the forces of selection.

Note also that all bases are subject to mutation, thereby producing an SNP, but SSR mutations only occur where a repetitive sequence already exists. (Also, the likelihood of mutation becomes significant only when the existing sequence has more than a few repeats, as the mechanism of mutation depends on this.)

A main question, though, is what is the significance of the variation? Are you interested in variation just because they mark individual genomes within a population, or are you interested in the variation of the phenotypes of individuals? Some genetic loci are irrelevant to phenotype and some can affect the phenotype to lesser or greater degree. SNPs can occur anywhere so can be presumed to have a "fair" chance of affecting phenotype. On the other hand SSRs can occur where only where repetitive sequences can occur (and persist where they survive selective forces). With SSRs, varying copy number in some contexts such as protein coding sequences, may have larger phenotype effects than an SNP, and thus be subject to stronger selection forces.

My view is that both SSRs and SNPs are important and it's not terribly productive to worry about general prominence or dominance of one over the other.


(1) Guichoux, Erwan & Lagache, Lelia & Wagner, Stefanie & Chaumeil, Philippe & Léger, Patrick & Lepais, Olivier & Lepoittevin, Camille & Malausa, Thibaut & Revardel, Emmanuelle & Salin, Franck & Petit, Rémy. (2011). Current trends in microsatellite genotyping. Molecular ecology resources. 11. 591-611. 10.1111/j.1755-0998.2011.03014.x.

  • $\begingroup$ The reason I'm interested is that I think that the nature and pacing of evolution is tied to the mechanism of mutation. I think there is relevance to modeling in population genetics. If I understand correctly, you are saying that there are far more SNPs in existence than copy-number variants of SSRs. On a per-base-pair basis, SSRs predominate, but, since they are fewer and further between, this might not mean more actual SSR mutation instances per generation,but it sounds like it it might favor SSRs by about an order of magnitude(4 orders of magnitude less in genome vs 5 orders more likely) $\endgroup$
    – johnnyb
    May 16, 2021 at 23:46

You must log in to answer this question.

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