Fst PCA

Matteo Fumagalli has been working on methods to estimate Fst and doing PCA/Covariance based on ANGSD output files.

The main documentation for this is found here: https://github.com/mfumagalli/ngsTools

Fst

1. Generate .saf files from each population using ANGSD SFS Estimation
   1. using a 2D-SFS as a prior, estimated using ngs2dSFS
   2. using marginal spectra as priors, estimated using realSFS

PCA

More information here: https://github.com/mfumagalli/ngsTools#ngscovar

cite

If you use these methods, you should cite the m. fumagalli paper http://www.ncbi.nlm.nih.gov/pubmed/23979584

New Fancy Method

  nRep <- 100
       nPop1 <- 24
       nPop2 <- 16
       cmd <- paste("msms -ms",nPop1+nPop2,nRep,"-t 930 -r 400 -I 2",nPop1,nPop2,"0 -g 1 9.70406 -n 1 2 -n 2 1 -ma x 0.0 0.0 x -ej 0.07142857 2 1  >msoutput.txt ",sep=" ")
       system(cmd)
       ##system("msms -ms 40 1 -t 930 -r 400 -I 2 20 20 0 -g 1 9.70406 -n 1 2 -n 2 1 -ma x 0.0 0.0 x -ej 0.07142857 2 1  >msoutput.txt  ")
 source("../R/readms.output.R")
  to2dSFS <- function(p1.d,p2.d,nPop1,nPop2)
        sapply(0:nPop1,function(x) table(factor(p2.d[p1.d==x],levels=0:nPop2)))

  source("../R/readms.output.R")
        a<- read.ms.output(file="msoutput.txt")
        
        p1.d <- unlist((sapply(a$gam,function(x) colSums(x[1:nPop1,]))))
        p2.d <- unlist((sapply(a$gam,function(x) colSums(x[-c(1:nPop1),]))))
        par(mfrow=c(1,2))
        barplot(table(p1.d))
        barplot(table(p2.d))
        sfs.2d <- t(sapply(0:nPop1,function(x) table(factor(p2.d[p1.d==x],levels=0:nPop2))))