ojwatson / mccoilr Goto Github PK

Perhaps we should not immediately remove loci with only homozygous calls in these lines?

	##remove sites and individuals with too much missing data
	simpleS = data[In_ind,]
	simpleS = simpleS[,In_site]
	
	##remove sites with P=0 or 1
	P=rep(NA, ncol(simpleS))
	for (j in (1:ncol(simpleS))){
		temp= simpleS[,j]
		P[j]= (sum(temp==1) + 0.5*sum(temp!=0 & temp!=1& temp!=-1))/sum(temp!=-1)
	}
	In = (P!=Inf & P!="NaN" & P!=0 & P!=1)
	simpleS2= simpleS[, In]
	
	select_pos =colnames(data)[In_site][In]
	select_ind = rownames(data)[In_ind]

Perhaps it might be better to throw a warning for the user to be aware they will be removed from following MCMC, or throw an error that then prompts users to specify an additional function argument in order to force their inclusion.

Add a parameter to control this, as it will save a lot of file memory:

int d = 2; // number of dp
// fprintf(V0, "%.*f", d, P[x]);

Both functions will fail with large data sets (>2000 by 110 SNP matrices roughly), as the containers for these are currently arrays and thus cause stack overflow issues.

Will be changed to dynamic memory allocation to overcome this.