Hello,
I am working on population genetics using STRs and, as the title explains, I tested for Hardy-Weinberg equilibrium using both Genepop and Cervus and applied Bonferroni correction. Even though some STRs are not much polymorphic, after the correction, I do not detect any significant deviations across the populations. However, the Fis values for some loci exceed 0.15 up to 0.33, which I understand is pretty high. The overall mean value is around 0.12. This means that my populations are inbreeding quite (?!) significantly, but shouldn't this reflect also in the HWE? I mean, if I was to defend such Fis values, could I reply that it is not an issue since I am still in HWE?
One last question, and I apologize if these are dumb questions but I am new to this kind of study, but why is no significance calculation ever performed on Fis values? Isn't such value also prone to random deviation (like my homozigosity is high just due to chance or sampling error)?
Thanks to anybody who will answer.
Best regards
Hello, it would indeed be expected to have HWE deviation for such Fis value.
A possible explanation could be related to a low sample size eventually?
There are possibilities to test significancy for Fis though permutations for instance (see Weir & Cockerham, 1984). Some software like Genetix do it.
Please note that positive Fis does not mean necessary inbreeding, it can also be related to population structure effect, or allele problems for instance.
Thank you for you answer Mr Leroy. I am working with 30 samples per population and I believe (and was told) it should be statistically significant.
By "allele problems" would you mean something like low polymorphism? My overall PIC (polymorphic information content) seems nice at around 0.5 and it does not really seem that the Fis values (calculated via FSTAT) show any particular deviation specifically at the less polymorphic loci (and by less polymorphic I mean both overall less polymorphic ones and also considering the actual number of alleles present in a specific population).
I am very befuddled.
30 samples look indeed fine to me.
When talking about allele problems I am referring more to null alleles, however since your average Fis is 0.12 across alleles, I won't think it is problem of specific markers.
Another question: did you have significant HW deficits for the markers with the largest Fis?
It indeed looks so! In many cases, high Fis values at a locus/ population combination, reflect border-line values of HWE at the same locus/pop combination. Usually ,thou, this is relevant only at a few loci (for example: pop1 at locus 1 displays high Fis value and low HWE P-value, but the other 10 loci look normal). There are some loci that seem overall more prone to this kind of bad result, but even getting rid of them and performing the calculation again does not solve the problem, meaning that this kind of high Fis values (and HWE deviation) are not due to particular markers (as you already pointed out), rather it is more some "bad locus/pop" specific combinations for each population. So, again, I am left with nothing but believing that such Fis is not high enough to make my pops deviate from HWE below the significance value.
I hope I explained crearly enough.
Thanks a lot for your help, Mr Leroy. I truly appreciate it.
Hi. Hope getting this right. I think It depends on how we analyze the data. If the Fis and HWE are estimated across markers, then some markers with excessive homozygosity (High Fis and deviate the HWE) are usually present. This could be due to null alleles as pointed out by Leroy. Also, these markers (or loci) could be under selection pressure (either natural selection or breeding selection). For instance, these loci may be related to fitness to survive, thus the individuals with new mutations on these loci do not survive. Thus, you would not detect any polymorphism on these loci in the population. Hope this will help, Michele.
Did you solve the problem Michele Marconcini ? I am having the same problem and think that, maybe, the type of reproduction strategy might be one of the answers.
Also Chee-Keng Teh could you please share some papers about the effects of selection in STR marker?
Thank you.
Hi, Sarah Vargas. I have enclosed one paper below for the presence of null allele of DNA markers (especially in microsatellites). You could run a test using MICROCHECKER program (please check the original paper below and look for the latest version) on all the assayed markers to estimate the frequency of null alleles. Instead of removing the affected markers, you could redesign the primers to avoid from the mutated annealing sites. If that is not feasible, I would suggest to replace these markers with other adjacent ones. Hope this would help! Cheers.
micro‐checker: software for identifying and correcting genotyping errors in microsatellite data
Article Null alleles are ubiquitous at microsatellite loci in the We...