I'm doing a population genetics study on a herbivore insect. Currently I'm looking for a way to search for associations between genotypes and the host plant preference. In other words I want to investigate whether there are different subpopulations targeting different plant species.
What do you think is the best way to perform such analysis? Would normal pairwise Fst tests be sufficient? or a single locus F-test ?
Hello,
Could you be more precise about your experiment, have you differentially sampled individuals on different host plants in the same locations and try to calculate Fst between "sub pops" sampled in different plants? I think F-stats should be helpfull but it will give you only information about the insect species, not the assocation with a putative preference for a host species, unless you just want to tell if there is differentiation between sub-pop sampled in different plant species, but the issue is it should be due to other things than host preference. You might want to consider multifactorial analysis to test for correlation between alleles at microsat markers and host plants. Model-based cluster analysis based on genetic data might also be helpful and you should find clusters that broadly correspond to host-preference if there is any. Finally you could also do a simple NJ tree and compare the clustering of individuals with host repartition.
I thank you both for your help .
Dr. Cubry , your explanations are very helpful
I have collected many samples from different geographical sites , and within each site samples were collected from different host plants , I calculated Fst Values to have an Idea about the geneflow between populations and to see if populations are differing regarding the geographical distance , until the now all I saw was that the geneflow rate is too high ( 0.01 < Fst < 0.1 ) , now I want to see if there are population divergence on different host plants in one site ..and I was wondering if the same test with the same principle would be sufficient
Well, that's not a so high Fst rate, but the important thing, more than the value, is the significance. What about it and Fis and Fit values? And the pairwise Fst between pops? A "classic" approach to adress gene flow based on geographical distance is to make an IBD (isolation by distance) test. It's basically a Mantel test between a matrix of genetic distance (Fst, or better, linearised Fst by Fst/(1-Fst)) and a geographical distance matrix. There are some very good reviews about Fst that have been recently made, one from Rousset in particular but can't remember the citation details. If the markers you used were microsatellites, you could also consider to use Rst, which are a Fst analogue taking into account the specific patterns of microsatellites (it's based on the number of repeats).
You can definitely try Fstat approach at a local scale considering several populations (one for each host species) or even at a global scale, again with a partition not on geographical populations but based on host preference. Like I've already said, a model-based analysis of genetic structure might be of help in this case.
I thank you very much Dr, Cubry for the explanations ...
now I will check the Fis and the Fit values .. when I was speaking about ( 0.01 < Fst < 0.1 ) is what I got from a pairwise Fst between pops, with a partition on geographical information ..
now I did the same and made the partitions based on host plant species ... I see higher Fst values than last time ...
I will figure out how to get Fis and Fit and how to do the isolation by distance analysis .. for my last analysis I used PowerMarker software V 3.25 and I don't think it does many other analysis ...
I will check other software and see them ,I also need to figure out how to get the significance values ... and I will be back then
I thank you very much for your contributions ... I do appreciate your help
- Badges
- Science topic
- Similar topics
- Medicine
- Physiology
- Biosignals
- Biomedical Signal Processing
- Statistical Analysis