# Read test dataset using the function read.table()
URL <-   #the location of the file on your computer, e.g. C://Rfiles//2pop.txt
data <- read.table("URL", header=T)
data[1:2,]

#Create genotype
# X: dataframe with fragment lengths x of differenct loci (columns)observed at N individuals (rows)
X <- data[,3:8]# if your data is differently organised, change the column assignment here.
X # check, if the columns you selected really give the loci data 
G <- X[,1]
    if(ncol(X)>1) for(i in 2:(ncol(X)))
    	{
    	G <- paste(G,X[,i],sep=".")
    	}
    G <- factor(G,levels=unique(G))

data$G <- G #attach G to data as column "G"
data[1,] # look at data file
attach(data) #attach data file

# This resampling function is needed to create a file containing no recurrent 
# multilocus genotypes
resample <- function(d=data)
{
    Trees <- split(1:nrow(d), G)
    Trees[[1]] # first data point

    a <- rep(NA,length(Trees))

    for(i in 1:length(Trees))
    {
    	a[i] <- Trees[[i]][1]
    	if(length(Trees[[i]]) > 1) 
    	{
	    a[i] <- sample(Trees[[i]], 1, replace=FALSE)
    	}
    }
    d[a,]
}

# In the following step, the object "data.res" without recurrent
# multilocus-genotypes is construced
data.res <- resample(data)
dim(data.res) # 23 13

# Make chi2 test and write chi2 and p-value into matrix
# X.matrix: p.value in the lower triangle, chi2 value in upper triangle
# Chi2: Sokal & Rohlf (2001)Biometry. Freeman, New York.
# T.matrix: p.value in the lower triangle, T in the upper triangle
# T is the Tschuproff contingency coefficient 
# T: Legendre & Legendre(1998)Numerical ecology, p.221
# V.matrix: p.value in the lower triangle, Cramer's V2 in the upper triangle
# V: Agresti, A. (1984)Analysis of ordinal categorical data. Wiley, New York, p. 23-24

# without recurrent multilocus genotypes 
Loci.res <- data.res[,3:8] # enter column numbers of your loci in "data.res"
L <- Loci.res #define this new if needed, e.g. if you enter an area or population
#
X.matrix <- matrix(NA, ncol(L), ncol(L), dimnames=list(names(L), names(L)))
T.matrix <- matrix(NA, ncol(L), ncol(L), dimnames=list(names(L), names(L)))
V.matrix <- matrix(NA, ncol(L), ncol(L), dimnames=list(names(L), names(L)))
for(i in 1:(ncol(L)-1))
	{
	for(j in (i+1):ncol(L))
		{
		chi <- c()
		tab <- table(L[,i], L[,j])
		chi <- chisq.test(x=tab, simulate.p.value=TRUE, B=2000)
		n <- sum(tab)
		d.f <- (ncol(tab)-1)*(nrow(tab)-1)
		X.2 <- chi$statistic
		T <- sqrt(X.2/(n*sqrt(d.f)))
		V.2 <- c()
		V.2 <- X.2/(n*(min(dim(tab)-1)))
		X.matrix[j,i] <- round(chi$p.value, 3)
		X.matrix[i,j] <- round(chi$statistic, 3)
		T.matrix[j,i] <- round(chi$p.value, 3)
		T.matrix[i,j] <- round(T ,3)
		V.matrix[j,i] <- round(chi$p.value, 3)
		V.matrix[i,j] <- round(V.2 ,3)
		}
	}
X.matrix
T.matrix
V.matrix # check the results. LPu03 and LPu20 are completely linked in the
# "2pop.txt" data example (p=0.000, V2=1.000). The other loci do not
# exhibit significant linkage disequilibrium.

# Save the object "V.matrix" as a text file for later importing into Excel
write.table(V.matrix, "C://Program Files//V.matrix.txt", sep=",")