Are Snps Frequencies Between Individuals Or Chromosomes

Question

[Pi]

Greetings

Following an interesting discussion on biostar here there are 2 views given whereby the allele frequency of a SNP is classed as between chromosomes and and the other is that the allele frequency of a SNP is classed as between individuals.

I personally thought the allele frequency was between individuals (e.g. a homozygous SNP counts as one occurence in the population and not two). I thought this was an interesting point and I was hoping for a resolution as I doubted many members of this forum would have spotted it on the tail-end of another question.

Yet again the definition of snp rears its ugly head.

==================

Edit: It seems I didn't word my question right. If the classical definition of a SNP is that is occurs in 1% of the population, does that mean that it occurs in 1% of the individuals in the population or 1% of the chromosomes in the population (i.e. the # of individuals x 2, assuming diploidy)

[lh3]

The answers below are all clear: it is 1% of the total chromosomes. Just forget individuals. Frequency is never defined that way.

[lh3]

The answers below are clear: it is 1% of the total number of chromosomes. Allele (SNP) frequency is never defined based on individuals.

[lh3]

Reply to "edit": the answers below are clear: it is 1% of the total number of chromosomes. Allele (SNP) frequency is never defined based on individuals.

[Pi]

okie dokie, karaoke. thanks

Answer 1

[Aaronquinlan]

Allele frequencies are based on alleles observed on chromosomes.

Human genotypes (e.g., A/G, A/A, G/G) are per diploid individual, whereas an allele is present on a single chromosome. A zygote is comprised of two haploid gametes (sperm and egg). Thus, at each polymorphic site, individuals are heterozygous (i.e., the sperm and egg had different alleles at that site) or homozygous (i.e., the sperm and egg had the same allele at that site).

As geneticists, we count the alleles present on chromosomes b/c chromosomes are the genetic material that segregate via sexual reproduction. For a biallelic site, the frequencies of the two alleles are denoted p and q. Assuming random mating, no selection, etc., the expected (binomial) genotype frequencies are thus p^2, 2pq, and q^2. I find it convenient to envision a big barrel filled with chromosomes, each having one of the two alleles. A genotype created from random mating is analogous to pulling two chromosomes (alleles) from the bag.

You state:

I personally thought the allele frequency was between individuals (e.g. a homozygous SNP counts as one occurence in the population and not two).

Here, you are referring to genotype frequencies. When observed genotype frequencies deviate from the binomial expectations, this is evidence for inbreeding, selection, CNVs, technology failures, etc.

Hartl and Clark have a brilliant text on the fundamentals of population genetics. I highly recommend it.

[Pi]

I've seen Hartl and Clark but it is was beyond my requirements for population genetics :)

Answer 2

[Casey Bergman]

Aaron is correct, the standard definition of allele frequency (SNP or otherwise) refers to the number of alleles of a particular type out of the total of 2N chromosomes at an autosomal locus in a diploid population of N individuals.

Where things can become a bit less clear is when sampling occurs in species (yeast, flies, plants, mice, etc.) where strains have been inbred or made isogenic prior to genotyping/sequencing to reduce heterozygosity (e.g. in order to allow full haplotype reconstruction, or to be able to pool multiple individuals from the same strain to get enough DNA for sampling). In these cases, the individual is treated as contributing a single chromosome to the sample, since inbreeding generates a diploid with two chromosomes that are derived from the same ancestral chromosome.

Answer 3

[Larry_Parnell]

I'll add a short bit here to bring into the discussion some of the different cases. These are allele frequencies in mitochondrial (MT) genomes and the haploid Y chromosome of human males. MT genomes are not uniform even in a single cell of a single person. Thus, for example, there is a mixture of genotypes in any one cell and this can make determinations of allele frequencies challenging. For the Y chromosome, remember that males have one copy, and one copy of X as well. This changes the 2N figure mentioned by others.

If you're interested in plant genomes, some are tetraploid or hexaploid.

[lh3]

For Y, divide by the number of individuals if not in PAR. For X, divide by 2*#females+#males if not in PAR. PAR is treated the same way as autosomes. Heteroplasmy in MT is very rare so far as I know. Unless you study heteroplasmy, treating MT as a single chromosome is sufficient most purposes.

[lh3]

The rule itself is simple: the denominator should be the total number of haplotypes. There may be the issue in the presence of CNVs, but in that case, SNPs are often questionable.