A confusion regarding haplotype

Question

[CY]

Say, allele A and B define haplotype 1. allele A, B and C define haplotype 2.

If we identify A, B and C at the same time, two situation may occur: 1) they all could exist in the same chromosome. 2) A and B in one chromosome and C in another chromosome.

This opens ambiguity between genotype 1* and 2. We can't just assume A, B and C are most likely in the same chromosome (statistically linked) because that way genotype 1 won't exist, right?

Any comments are appreciated!

[Kevin Blighe]

I think that the OP is referring to 'phasing'.

OP: you meant that they cannot exist on the same maternal or paternal chromosome right?

If you identify A, B, and C using standard NGS technology, you will find it very very difficult to determine whether they have occurred on the same maternal or paternal chromosome. Phasing information will not exist in the VCF.

You could yes determine phasing information via the latest 10x technology by using special adaptors that allow you to determine phasing (I recently did a whole genome experiment this way). Some genotype arrays may also allow it.

OP: you are correct in your assumption, though, in that we cannot know whether they have occurred on the same maternal or paternal chromosome, UNLESS they happen to all be homozygotic calls.

[shoujun.gu]

Now I may understand what the post means...

[christopher medway]

Do you mean "allele A and B define haplotype 1. allele A, B and C define haplotype 2"?

[CY]

Yes, sorry for the confusion

[WouterDeCoster]

See also Do multiple SNPs exist in same chromosome to be called allele?

Answer 1

[WouterDeCoster]

A haplotype is a block of alleles of SNPs that are frequently inherited together as one "set" - i.e. it's sufficiently close to each other that no recombination occurs (frequently) between those (which would "break" the haplotype). As such those are relatively stable between generations (although obviously recombination and mutation happens).

Two chromosomes are independently inherited (randomly separated), and as such, no haplotype can exist "across" chromosomes.

Answer 2

[christopher medway]

If you have genotyped three SNPs within a single sample, you would not be able to infer the haplotypes without additional information - so you are basically correct. If you have a large number of samples genotyped at these three SNPs, you can begin to ascribe probabilities to each haplotype. For example, if you notice that the three SNPs always have the same genotype within a sample, then this is strong evidence that the three alleles are in phase; A-B-C [Haplotype 2]

Answer 3

[CY]

This is maybe quick and dirty solution:

In case we found all three variants and encounter two possible haplotypes, we can check variant C in dbsnp. It is most likely a minor allele. If it is frequency is reasonably low, we can assume *2 is the most likely answer. What do you think?

[christopher medway]

So you are starting with the assumption that these three variants only form two haplotypes; i) A-B-X and ii) A-B-C (X is the reference allele) - which I will work with here. You can use resources such as the HapMap project to check the relative frequencies of each haplotype in your population of interest (European ... ). If this resource tells you that A-B-C are never seen in phase in 300 European samples, then you have good evidence that A-B-X is the haplotype. Your suggested method, using minor allele frequency of allele C does not take into account linkage disequilibrium. I think you should read WouterDeCoster's post again, and the other links that are up here.