I have three population to compare their allele frequency. Based on the previous AFLP fingerprint, we know B is derived from A. A and B are closer to each other compared to C, but B shows different phenotype from A and C.
We did RNAseq on those three populations and identified the differential expression genes in each pairwise comparison. After that, I tried to do some SNPs analysis among three races, assuming the mutation may happen on protein coding region which not necessarily results in differential expression. So when identifying the SNPs, I removed all the differential expression genes in each pairwise comparisons and focused on SNP changes from those non differential expressed gene only. Surprisingly, the allele frequency changes (see distribution image: https://www.dropbox.com/s/z793h29xja9x21f/Untitled-9.png?dl=0) are more dramatic between B and A than between A and C, which is controvert to our AFLP experiment.
I am a naive on SNP analysis. Does allele frequency changes among different population tell you how distant those populations are? How should I explain the discrepancy between AFLP results and allele frequency change result?
We need some more info to help you:
1) How did you compute allele frequency? Is it a) the frequency of the reads carrying the allele in ONE individual or b) the frequency of the chromosomes carrying the allele in a population?
I would not recommend the use of option a), since it can be heavily influenced by gene expression patterns (for which I would not surprised to see B as the outlier, as in the phenotype)
Also, why did you use RNAseq if you want to compare allele frequencies? Couldn't you just type some DNA marker?
If you want to infer genetic distances you might reconstruct the sequences of the genes and perform a classic phylogenetic analysis on the sequences.