Post date: Jul 30, 2015 5:2:23 PM
We wanted to verify that MAFs were not strong drivers of patterns of Fst at the 20kb window-level for the T. cristinae genome data (A vs. C survivors from the experiment and the four natural population pairs).
The results from the experimetn are in /labs/evolution/projects/timema_speciation/timema_wgexperiment/popgen/. I used the script calcNeiPiMaf.pl to calculate mean MAF (between pops and for 20 kb windows) and Fst (also for 20 kb windows). The output is in mafnei20kv3_AsurvCsurv.txt, and the R code I used to plot the results is here:
x<-read.table("mafnei20kv3_AsurvCsurv.txt",header=FALSE)
pdf("expMafFst.pdf",width=6,height=6)
par(mar=c(5,5,0.5,0.5))
plot(x[,4],x[,5],pch=20,xlab="20kb window mean MAF",ylab="20kb window Fst",cex.lab=1.4)
text(0.4,0.04,"r = 0.002")
text(0.4,0.037,"p = 0.758")
dev.off()
MAF and Fst are essentially independent (r = 0.002, p = 0.758).
Victor tested the natural populations. He also found no real correlation, but he used a (possibly inappropriate) 1% MAF cutoff. I doubt it will matter, but I sent him the correct files (the ones I used for Fst calculations) to check this out.