# Statistician confused about exact SNP data type

I'm from a statistical (not biological) background, and I'm very confused about what exactly is the "data" associated with a SNP. Below, I'll explain things as best I can--please correct me if any point is wrong or otherwise unclear. Assume we're working with humans, since the following will change depending on the species. (Assume all cases of interest are biallelic.)

When a locus has an allele on least one chromosome, the measurement AA, Aa, or aa is called a SNP (pronounced "snip"). Therefore, using statistical language, SNPs are three level factors--this is the data associated with a SNP. As said earlier, if the factor has 1 level (ie it's only aa or only AA) then it's not a SNP.

Sometimes for humans it seems like people say a SNP is actually just one allele (so a two level factor, a or A). How is this possible? Do people sometimes only measure one allele instead of both to get the genotype?

• Could you provide the source where they refer only to one allele? Commented Mar 7, 2021 at 20:06
• @Hachiloni There is no one source, but this is an example: journals.plos.org/plosone/article?id=10.1371/… . See the paragraph which begins "Second, we compare three SNP codings..." Commented Mar 7, 2021 at 20:10
• You have laid out what your knowledge of SNPs in an admirable fashion, except for one important thing. You do not state what SNP stands for. The name describes the concept: single nucleotide polymorphism. A polymorphism is literally a difference in shape (here a difference in the physical form) of an object. The object is the DNA sequence of a gene, and the difference is in a single nucleotide. It is not defined by whether the alleles are Aa, AA or aa — just by the fact that at this position different bases are found (generally 2, but could be 3 or 4). Do what you will with your statistics. Commented Mar 7, 2021 at 22:08
• Thank you, @David. That's a very helpful point--I ignored where the term "SNP" came from completely. (I've edited the answer to address that I'm focused on the biallelic case.) I'm not focused here on what to do with the data (i.e. what statistics to do), just what the data itself looks like. I believe this is a biological question since it's certainly not statistical. If the data was "a" or "A", would you think that's SNP data? Or must the data be "aa","Aa", or "AA" to be SNP data? Commented Mar 7, 2021 at 22:26
• I don’t know. But surely there is a vast literature on this. I read (skim) papers in Nature on large comparative genome projects which refer to the incidence of SNPs. There was one in February. The impression I get is they are just looking at whether different bases are found at particular points and how many there are in the genomes they have sampled. You would really have to find the references to their methods and check yourself (unless someone else here knows). Commented Mar 7, 2021 at 23:48

SNP is not a specific data type but rather a biological phenomenon. The abbreviation "Single Nucleotide Polymorphism" only means there is a variability (between individuals) in a single letter in a specific position in DNA sequence. It depends on the speaker how they decide to describe the variability; wether they choose to focus on the possible sequences and say for example "There is only A or T in this position" or they focus on possible human genotypes and spells out all combinations (AA, AT, TT).

What constitutes as SNP data depends entirely on research question, used organism (diploid/haploid/polyploid) and experimental design.

Article you link in your comment uses oat lines and not individuals from outbreeding population. Oat like arabidopsis is naturaly highly self polinating so creating highly homozygous inbred lines is relatively easy. In this case it is reasonable for the researches to expect only AA or aa genotypes in each line.

Edit: Since you asked for more examples where the researcher chooses to focus on alleles instead of genotypes. Many SNP genotypes distribution fit Hardy-Weinberg equations. If that is the case, the frequency of genotypes can be calculated from an allele frequency in population. The researcher might ask: "Is the frequency of allele a different in population/group 1 from population/group 2?" They collect data from indivuduals in group 1 (AA, AA, Aa, Aa, aa) and group 2 (Aa, Aa, aa, aa, aa). They decide to treat each allele as independent measurement (not dependent on individual it comes from) and poll the data: group 1 (AAAAAAaaaa), group 2 (AAaaaaaaaa) and then they do simple binomial test to answer their question.

Note: Even if Hardy-Weinberg equilibrium is not met, the questions about of allele frequency are is still valid biological questions (with slightly different use/interpretation down the line).

• Thanks. Could you provide other examples where it's reasonable to only focus on the value of one allele (when ploidy is at least 2) to constitute a SNP? Your argument for the linked paper seems to be that researchers who care about the genotype can only measure one allele, which is interesting but circumvents the core problem I'm having. Commented Mar 8, 2021 at 6:25
• Thanks for including the interesting edit. Not to bleed you dry here, but unfortunately due to my limited biological background, any allusion you make it completely lost on me. Could you clarify with examples what you're describing in the last sentence on when HW is not met? Commented Mar 8, 2021 at 15:23
• I can refer you to Khan academy on HW principle. HW has several asumptions any of which can be not true in a real population. What I ment to say is that even in this situation it is still meaningful to talk about allele frequency. For example when a species is highly self polinating (like oat) resulting in more homozygots than expected from HW, we can still ask: is one allele more frequent in eg. high altitude population? Commented Mar 9, 2021 at 5:53

I think the prevuious answers provide lots of valuable and relevant information, but let me add additional prospective:

• As noted, SNP is a single nucleotide polumorphism, whereas notation AA, Aa, aa, etc. may refer to many other types of mutation: e.g., these could be whole genes, differing on many locations. Another common type of mutation is insertions/deletions, where one gene may lack whole segments (rather than contain different nucleotides).
• SNPs are used in particular contexts, where the change of genome is insignificant: e.g., when treating highly conserving human genome or when treating a variable genome on a short time scale, where only few changes happen. In this case we take the most frequent/initial genome as the reference and describe deviations from it as SNPs ("snips"). This is not always, possible - e.g., when multiple alleles are present (a single nucleotide can be in four states, A, C, G, T and all of these may have to be accounted for, e.g., when analyzing the evolution of viruses).
• Depending on the problem, one may want to discuss population genetics of an organism in terms of genotypes or in terms of allels - even when a single organism carries multiple allele (e.g., two copies in a diploid organism). AA, Aa, aa are genotypes, whereas A, a are alleles. SNP is a change in allele rather than in a genotype. (But I do admit that biological language is often unprecise.)
• Wow, thank you. This is incredibly helpful and fills in a lot of missing gaps for me. Commented Mar 8, 2021 at 15:26
• @user257566 I also come from non-biological background, so I probably encountered similar issues in understanding these things. For population genetics I recommend also the Gillespie's little book. It is somewhat sketchy, but might be easier to work through than real pop gen stuff: amazon.com/Population-Genetics-John-H-Gillespie/dp/0801880092 Commented Mar 8, 2021 at 15:32

Let me give you a back of the envelope calculation. The size of the human genome is 3 billon base pairs (BP). There are about 30 thousand genes in the human genome. Therefore, the size of a gene would be around 100 thousand BP (DNA does have non-coding regions, regulatory regions etc. Also a gene vary in size from a few hundred BP to more than 2 million. For the sake of simplicity, we can stick to this calculation). On the other hand, there are 5 million SNPs in human genome. Therefore, There would be a SNP in every 600 BP.

So, we can say with certainty that there would be genes that have more than one SNP. My suggestion would be, as @BagiM pointed out, do not confuse microscopic concepts with macroscopic ones.

• Thank you for sharing your perspective. By "a or A" I meant an allele at a single locus, by "aa or Aa or AA", I meant a pair of alleles which constitute a genotype (in a human). Could you clarify what you mean by switching between microscopic and macroscopic? Commented Mar 8, 2021 at 15:25
• SNP is a microscopic concept and Allele is a macroscopic concept. Allele is a very broad term and the molecular basis of this term could be numerous (@Vadim discussed it in his answer). The term was coined by Mendel himself. That too a long time before DNA was discovered. "a SNP is actually just one allele" -- is a wrong statement. There could be multiple SNPs inside an allele.
– Noob
Commented Mar 8, 2021 at 15:57
• Thank you for clarifying my terminology--that's very helpful. I'm referring specifically to an allele as a variant in one base pair, and a genotype (for humans) as a pair of such alleles still at one base pair but now across two chromosomes. So my question is specifically about which one (between these two microscopic things) is used to represent SNPs. Commented Mar 8, 2021 at 18:13
• None. SNPs are not defined in terms two chromosomes. Huge collection of DNA sequences are compared and specific positions on the sequence are found to show this polymorphism (which we call SNPs) while other regions of the sequence remain relatively conserved.
– Noob
Commented Mar 8, 2021 at 23:56
• Ok, by none do you mean the former? The single loci? Commented Mar 9, 2021 at 1:29

As said earlier, if the factor has 1 level (ie it's only aa or only AA) then it's not a SNP.

If there is any variation in that base in your population, then it is a SNP, even if you are not looking at a heterozygote.

Anyway, if you are talking about a SNP, people will generally refer to it by its base, not be the name of the allele it generates. (And actually lots of time, the alleles are not named as simply as A and a. That notation is helpful when dealing with perfectly Mendelian middle school genetics homework problems, but in the real world, most data doesn't fit that paradigm.

• Thanks. That quoted passage is in fact saying if there's no variation (ie if its only aa or only AA that have been observed). Further, it's been a while since I've been to middle school so it's helpful to know what's covered there for context. However, I'm not talking about how to name SNPs (personally, I'm familiar with rs numbers for that.) Instead. I'm talking about what values they take. Names for things are helpful for me here than the concept they represent. Commented Mar 8, 2021 at 19:04
• Why would you have notation for different alleles if there was no variation in any population? Commented Mar 8, 2021 at 19:07