introgress issues with individuals

[GabrieleNocchi]

Hi there,

I am having an hard time understanding the last few steps of the introgress tutorial. I am running introgress with all the same indidivuals in the previous maxent step, as a test. My genind object has got 380 individuals. I create the files with:

genind2introgress(gen = g_ind, p1 = "parent1", p2 = "parent2", admix = "admixed", prefix=file.path(plotDIR, "eatt"))

Here parent1 parent2 and admixed are my populations defined in @pop of the genind.

Now if I sum the ncol of the resulting input files, they sum to 382. The number of columns of each created input file seem to correspond to the number of individuals in my parent1 and parent2 populations, except the admix file which has 2 extra columns. I don t understand the admix file, as it seems to have rows = SNPs as I have 310 SNPs in my test and 310 rows, but for some reason there are 310 indidividual IDs in the second column which I have no idea what they are. My admixed individuals are 224. The total number of col of admixed input file is 224 + 2.

Then I run introgress, and results from the introgress run are given for 226 indidivuals. This number does not correspond to the admixed indidivuals number (which is 224), does not correspond to the individuals in column 2 of eatt_admix.txt (310 individuals ) and therefore I am unable to go past the subset function:

rasterPoint.list.subset <-
lapply(rasterPoint.list,
subsetIndividuals,
file.path(dataDIR, "inputFiles", "eatt_inds.txt"))

I have no idea which individuals I need to subset from rasterPoint.list (which has all the indidivuals in the genind) to match the 226 in the introgress output, and therefore cannot move to the final function clinesXenvironment.

I think I am missing something.....I am wondering maybe some issues in reading my genind file in how I specified my populations?

[GabrieleNocchi]

Indeed. Some big issues in either how vcfR reads in the VCF and then converts it to genind or in your conversion from genind to introgress . I suspect in vcfR. Everything seems good if I start from a gz VCF rather than uncompressed....

[GabrieleNocchi]

I think I have found the issue, it is in the genind2introgress function of your package. Line 86 of your code (of genind2introgress.R) , for the admixed input file you do this (not sure why):
admix.data <- rbind(pop, row.names(admix.data), admix.data)

This gets messed up when you transpose the data upon creating the file:

write.table(data.table::transpose(admix.data),
              file=paste0(prefix, "_admix.txt"),
              sep="\t", col.names=FALSE, row.names=FALSE,
              quote = FALSE)

In addition, your example data admixed input file in the dryad do not have these 2 extra rows/columns, it looks exactly the same as the p1 and p2 input files (as it should be). I am not sure why/when that faulty line in the code was added. Works without it.

Cheers,
G.

[GabrieleNocchi]

This is how I modified the code of the genind2introgress function:

##########################################################################################################################################

#' Generate inputs for Introgress from genind object
#'
#' This function parses a genind object to create input data frames for
#' running genomic cline analysis in the Introgress R package
#'
#' @param gen Genind object containing data for parental and admixed populations
#' @param p1 Character or vector containing population names for the p1 group
#' @param p2 Character or vector containing population names for the p2 group
#' @param admix Character or vector containing population names for the admixed individuals
#' @param missingPerInd Proportion of missing genotypes allowed to retain an individual
#' @param missingPerLoc Proportion of missing genotypes allowed to retain a locus
#' @param subset An optional parameter defining a number of loci to randomly retain
#' @param drop Boolean value defining whether or not to remove monomorphic loci
#' @param prefix Prefix for output Introgress files (default="introgress")
#' @export
#' @examples
#' introgress_input <- genind2introgress(genind, p1=c("PopA", "PopB"),
#'                                         p2="PopC", admix=c("PopD1", "PopD2"))
#'
#' introgress_input <- genind2introgress(genind, p1=c("PopA", "PopB"),
#'                                         p2="PopC", admix=c("PopD1", "PopD2"),
#'                                         missingPerInd=0.75, missingPerLoc=0.25,
#'                                         drop=TRUE)
#'
#' introgress_input <- genind2introgress(genind, p1=c("PopA", "PopB"),
#'                                         p2="PopC", admix=c("PopD1", "PopD2"),
#'                                         subset=500)
genind2introgress <- function(
	gen,
	p1,
	p2,
	admix,
	missingPerLoc=0.5,
	missingPerInd=0.5,
	subset=NULL,
	drop=TRUE,
	prefix="introgress"
){

	ret<-list()

	all_pops<-c(p1, p2, admix)

	missingPerLoc = 1.0 - missingPerLoc
	missingPerInd = 1.0 - missingPerInd
	# drop individuals not in selected pops
	sub<-gen[adegenet::pop(gen) %in% all_pops,]

	# drop loci having more than missingPerLoc missing data
	sub<-filter_missingByLoc(sub, prop=missingPerLoc)

	# drop individuals having more than missingPerInd missing data
	# NOTE: Also dropping loci that become monomorphic
	sub<-filter_missingByInd(sub, prop=missingPerInd, drop=drop)
  #print(sub)
	# if necessary, randomly subset loci to the number requested
	if (!is.null(subset)){
		sub<-filter_randomSubsetLoci(sub, sample=subset)
	}
  #print(sub)
	# begin parsing remaining data to create inputs for introgress
	# set up loci information table ()
	loci.data<-data.frame(cbind(c(names(sub$loc.n.all)), sub$type), stringsAsFactors = FALSE)
	colnames(loci.data)<-c("locus", "type")
	loci.data[loci.data["type"]=="codom", "type"]<-"C"
	loci.data[loci.data["type"]=="dom", "type"]<-"D"

	# genotype table for p1
	p1_genind<-sub[adegenet::pop(sub) %in% p1,]
	p1.data<-adegenet::genind2df(p1_genind, sep="/")
	p1.data<-subset(p1.data, select=-c(pop))
	p1.data[is.na(p1.data)] <- "NA/NA"

	# genotype table for p2
	p2_genind<-sub[adegenet::pop(sub) %in% p2,]
	p2.data<-adegenet::genind2df(p2_genind, sep="/")
	p2.data<-subset(p2.data, select=-c(pop))
	p2.data[is.na(p2.data)] <- "NA/NA"

	# genotype table for admix population
	admix_genind<-sub[adegenet::pop(sub) %in% admix,]
	admix.data<-adegenet::genind2df(admix_genind, sep="/")
	pop<-admix.data$pop
	admix.data<-subset(admix.data, select=-c(pop))
	admix.data[is.na(admix.data)] <- "NA/NA"
	#admix.data <- rbind(pop, row.names(admix.data), admix.data)

	# write output files
	print(paste0("Writing outputs with prefix: ", prefix))
	if (file.exists(paste0(prefix, "_p1data.txt"))) {
	  # Delete file if it exists
	  file.remove(paste0(prefix, "_p1data.txt"))
	}
	write.table(data.table::transpose(p1.data),
	            file=paste0(prefix, "_p1data.txt"),
	            sep="\t", col.names=FALSE, row.names=FALSE,
	            quote = FALSE)

	if (file.exists(paste0(prefix, "_p2data.txt"))) {
	  # Delete file if it exists
	  file.remove(paste0(prefix, "_p2data.txt"))
	}
	write.table(data.table::transpose(p2.data),
	            file=paste0(prefix, "_p2data.txt"),
	            sep="\t", col.names=FALSE, row.names=FALSE,
	            quote = FALSE)

	if (file.exists(paste0(prefix, "_loci.txt"))) {
	  # Delete file if it exists
	  file.remove(paste0(prefix, "_loci.txt"))
	}
	#l<-as.data.frame(t(loci.data))
  write.table(loci.data, file=paste0(prefix, "_loci.txt"),
            sep="\t", col.names=TRUE, row.names=FALSE,
            quote = FALSE)

  if (file.exists(paste0(prefix, "_admix.txt"))) {
    # Delete file if it exists
    file.remove(paste0(prefix, "_admix.txt"))
  }
  write.table(data.table::transpose(admix.data),
              file=paste0(prefix, "_admix.txt"),
              sep="\t", col.names=FALSE, row.names=FALSE,
              quote = FALSE)
}

###############################################
#' Utility function to filter missing data by locus.
#' @param gen DataFrame Genotypes to filter.
#' @param prop Float Missing data threshold for which to filter loci.
#' @return DataFrame A filtered DataFrame where loci with missing data > prop are removed.
#' @noRd
filter_missingByLoc<-function(gen, prop=0.5){
  propTyped<-adegenet::propTyped(gen, by="loc")
	#print(missing)
	#print(missing>prop)
	#print(genind[loc=c(missing>prop)])
	return(gen[loc=c(propTyped>prop)])
}

#' Utility function to filter missing data by individual.
#' @param gen DataFrame Genotypes to filter.
#' @param prop Float Proportion of missing data threshold.
#' @param drop Boolean Whether to drop monomorphic loci.
#' @return DataFrame Individuals with missing data > prop filtered out.
#' @noRd
filter_missingByInd<-function(gen, prop=0.5, drop=TRUE){
	propTyped<-adegenet::propTyped(gen, by="ind")
	#print(propTyped)
	#print(prop)
	#print(propTyped)
	#print(propTyped>prop)
	return(gen[c(propTyped>prop), drop=drop])
}

#' Utility function to randomly subset loci.
#' @param gen DataFrame Genotypes to randomly subset.
#' @param sample Integer Number of loci to subset.
#' @return DataFrame Random subset of genotypes.
#' @noRd
filter_randomSubsetLoci<-function(gen, sample){
	loc<-adegenet::nLoc(gen)
	if (sample < loc){
		locs <- sort(sample(1:loc, sample, replace=FALSE))
		return(gen[loc=c(locs)])
	}
}
###########################################################################################################################################################################