Multiple heterozygous groups

CramCycle

Looking at some clustering on a SNP ID but instead of the usual three groups (2 homozygous, 1 heterozygous), I am seeing what is looking like 2 individual heterozygous clusters. I vaguely remember a postdoc saying something about primer dimers forming for various reasons but I cannot remember. Are there simple explanations? how would you ID which one is genuine? or is it a no call for both?




I made the calls on both BTW

Kevster

It is possible to have more than just the 'A' and 'B' allele calls at the same position, and even some variants listed in dbSNP will show, as an example, [C/T/G], i.e., the variant can have 3 possible values.

What position are you genotyping?

Edit: of course, if this was an indel, it could have multiple possible values depending on the sensitivity of the technology that you're using. For example, I analyse sequencing data a lot and an indel like AAATGGC could be called in multiple different ways by the variant callers.

CramCycle

The markers used are only designed to amplify one of the two possible alleles so it would not matter if there were more available.

Used KASP for the above, so it has one forward primer for one allelle, another forward primer for the other allele. indels would mean that the KASP would likely not work at all.

If you were using something like TAQ probes you would have the same thing as they only target one of the two known alleles.

I think I figured out a possibility in that the samples quality may have varied between samples, so we get mild PCR inhibition on a plate and that moves its cluster position but so far that's my only logical theory but no one else seems to have commented on it anywhere else before.

Kevster

I think that I get what you mean; however, a normalisation method performed on the raw data (such as RMA, GC-RMA, quantile, or others) should eradicate the strange plot that was produced. This would have the effect of merging those 2 different heterozygote groups into 1.

I don't know the true nature of the experiment that was conducted though! I typically do experiments where tens of thousands or even millions of SNPs are genotyped at once.

Kevin

CramCycle

Kevster wrote: »

I don't know the true nature of the experiment that was conducted though! I typically do experiments where tens of thousands or even millions of SNPs are genotyped at once.

I am in the opposite end of things, a small number of SNPs (100 - 200) against thousands of samples. I have looked at data from Affychips and the like so I know what your referring to (sort of). I don't normalise the DNA (too costly and time consuming) which leads to issues with the KASP so I have to re-cycle the chemistry a few times and follow the movement. i know from sending off samples for an affy chip that they were very serious about the concentration.

Only normalisation we have here is the software does background ROX so you can be sure that you added in the right amount of master mix and it will try and adjust the position if the ROX concentration is off.