R/dist.xyStats.R

In kullrich/distIUPAC: calculates IUPAC distances on Biostrings DNAStringSet and BStringSet objects

#' @title dist.xyStats
#' @name dist.xyStats
#' @description This function calculates \code{distIUPAC} based distances
#' comparing two populations (x: receiver; y: donor).
#' @importFrom stats as.dist sd setNames
#' @param tmpSEQ \code{DNAStringSet} [mandatory]
#' @param x.pos population X positions [mandatory]
#' @param y.pos population Y positions [mandatory]
#' @param x.name population X name [default: "x"]
#' @param y.name population Y name [default: "y"]
#' @param dist distance to use [default: IUPAC] or choose one model as in
#' \link[ape]{dist.dna} [default: "IUPAC"]
#' @param ncores number of parallel cores to process pairwise distance
#' calculation [default: 1] see \link[distIUPAC]{rcpp_distIUPAC} [default: 1]
#' @seealso \code{\link[distIUPAC]{xyStats}},
#' \code{\link[distIUPAC]{distIUPAC}}, \code{\link[distIUPAC]{rcpp_distIUPAC}}
#' @examples
#' ##Use the 'xyStats' function to handle population assignment automatically
#' 
#' ##load sequence data
#' data("MySequences", package="distIUPAC")
#' CAS.pos<-5:34
#' AFG.pos<-82:87
#' 
#' ##Here, one needs to consider the changed x and y positions due
#' ##to sub-sampling.
#' CAS.pos_<-seq_along(CAS.pos)
#' AFG.pos_<-seq(from=length(CAS.pos_)+1, to=length(c(CAS.pos,AFG.pos)))
#' 
#' ##pairwise deletion
#' CAS_AFG.pairwiseDeletion.dist<-distIUPAC(
#'   as.character(MySequences[c(CAS.pos,AFG.pos)]))
#' CAS_AFG.pairwiseDeletion.xyStats<-dist.xyStats(
#'   CAS_AFG.pairwiseDeletion.dist, 
#'   x.pos=CAS.pos_, y.pos=AFG.pos_,
#'   x.name="CAS", y.name="AFG")
#' 
#' ##global deletion
#' CAS_AFG.globalDeletion.dist<-distIUPAC(
#'   as.character(globalDeletion(MySequences[c(CAS.pos,AFG.pos)])))
#'   CAS_AFG.globalDeletion.xyStats<-dist.xyStats(CAS_AFG.globalDeletion.dist,
#'   x.pos=CAS.pos_, y.pos=AFG.pos_, x.name="CAS", y.name="AFG")
#' 
#' ##compare results
#' rbind(CAS_AFG.pairwiseDeletion.xyStats, CAS_AFG.globalDeletion.xyStats)
#' @export dist.xyStats
#' @author Kristian K Ullrich
#dist.xyStats<-function(dIUPAC, x.pos, y.pos, x.name="x", y.name="y"){
dist.xyStats<-function(tmpSEQ, x.pos, y.pos,
  x.name="x", y.name="y", dist="IUPAC", ncores=1){
    options(scipen=22)
    if(dist=="IUPAC"){
        dIUPAC<-rcpp_distIUPAC(as.character(tmpSEQ), ncores=ncores)
    }
    if(dist!="IUPAC"){
        dIUPAC<-setNames( list(
          dist.dna(as.DNAbin(DNAMultipleAlignment(tmpSEQ)),
            model=dist, as.matrix=TRUE, pairwise.deletion=TRUE),
          pairwiseDeletion(as.character(tmpSEQ))$sitesUsed
          ), c("distIUPAC", "sitesUsed") )
    }
    XNAME<-x.name
    YNAME<-y.name
    dMean.x<-NA
    dSd.x<-NA
    dSites.x<-NA
    dMin.x<-NA
    dMax.x<-NA
    dMean.y<-NA
    dSd.y<-NA
    dSites.y<-NA
    dMin.y<-NA
    dMax.y<-NA
    dMean.xy<-NA
    dMean_.xy<-NA
    dSd.xy<-NA
    dSites.xy<-NA
    dMin.xy<-NA
    dMax.xy<-NA
    dTotal.xy<-NA
    dSweighted.xy<-NA
    Fst.xy<-NA
    dRelative.xy<-NA
    OUT<-list(XNAME, YNAME,
      dMean.x, dSd.x, dSites.x, dMin.x, dMax.x,
      dMean.y, dSd.y, dSites.y, dMin.y, dMax.y,
      dMean.xy, dMean_.xy, dSd.xy, dSites.xy, dMin.xy, dMax.xy,
      dTotal.xy, dSweighted.xy, Fst.xy, dRelative.xy)
    names(OUT)<-c("XNAME", "YNAME",
      "dMean.x", "dSd.x", "dSites.x", "dMin.x", "dMax.x",
      "dMean.y", "dSd.y", "dSites.y", "dMin.y", "dMax.y",
      "dMean.xy", "dMean_.xy", "dSd.xy", "dSites.xy", "dMin.xy", "dMax.xy",
      "dTotal.xy", "dSweighted.xy", "Fst.xy", "dRelative.xy")
    #x
    if(length(x.pos)==1){
        OUT$dMean.x<-mean(dIUPAC$distIUPAC[x.pos, x.pos], na.rm=TRUE)
        OUT$dSd.x<-sd(dIUPAC$distIUPAC[x.pos, x.pos], na.rm=TRUE)
        OUT$dSites.x<-mean(dIUPAC$sitesUsed[x.pos, x.pos], na.rm=TRUE)
        OUT$dMin.x<-min(dIUPAC$distIUPAC[x.pos, x.pos], na.rm=TRUE)
        OUT$dMax.x<-max(dIUPAC$distIUPAC[x.pos, x.pos], na.rm=TRUE)
    }
    else{
        OUT$dMean.x<-mean(as.dist(dIUPAC$distIUPAC[x.pos, x.pos]), na.rm=TRUE)
        OUT$dSd.x<-sd(as.dist(dIUPAC$distIUPAC[x.pos, x.pos]), na.rm=TRUE)
        OUT$dSites.x<-mean(as.dist(dIUPAC$sitesUsed[x.pos, x.pos]), na.rm=TRUE)
        OUT$dMin.x<-min(as.dist(dIUPAC$distIUPAC[x.pos, x.pos]), na.rm=TRUE)
        OUT$dMax.x<-max(as.dist(dIUPAC$distIUPAC[x.pos, x.pos]), na.rm=TRUE)
    }
    #y
    if(length(y.pos)==1){
        OUT$dMean.y<-mean(dIUPAC$distIUPAC[y.pos, y.pos], na.rm=TRUE)
        OUT$dSd.y<-sd(dIUPAC$distIUPAC[y.pos, y.pos], na.rm=TRUE)
        OUT$dSites.y<-mean(dIUPAC$sitesUsed[y.pos, y.pos], na.rm=TRUE)
        OUT$dMin.y<-min(dIUPAC$distIUPAC[y.pos, y.pos], na.rm=TRUE)
        OUT$dMax.y<-max(dIUPAC$distIUPAC[y.pos, y.pos], na.rm=TRUE)
    }
    else{
        OUT$dMean.y<-mean(as.dist(dIUPAC$distIUPAC[y.pos, y.pos]), na.rm=TRUE)
        OUT$dSd.y<-sd(as.dist(dIUPAC$distIUPAC[y.pos, y.pos]), na.rm=TRUE)
        OUT$dSites.y<-mean(as.dist(dIUPAC$sitesUsed[y.pos, y.pos]), na.rm=TRUE)
        OUT$dMin.y<-min(as.dist(dIUPAC$distIUPAC[y.pos, y.pos]), na.rm=TRUE)
        OUT$dMax.y<-max(as.dist(dIUPAC$distIUPAC[y.pos, y.pos]), na.rm=TRUE)
    }
    #xy
    OUT$dMean.xy<-mean(dIUPAC$distIUPAC[x.pos, y.pos], na.rm=TRUE)
    OUT$dMean_.xy<-OUT$dMean.xy - (OUT$dMean.x/2) - (OUT$dMean.y/2)
    OUT$dSd.xy<-sd(dIUPAC$distIUPAC[x.pos, y.pos], na.rm=TRUE)
    OUT$dSites.xy<-mean(dIUPAC$sitesUsed[x.pos, y.pos], na.rm=TRUE)
    OUT$dMin.xy<-min(dIUPAC$distIUPAC[x.pos, y.pos], na.rm=TRUE)
    OUT$dMax.xy<-max(dIUPAC$distIUPAC[x.pos, y.pos], na.rm=TRUE)
    OUT$dTotal.xy<-mean(as.dist(dIUPAC$distIUPAC[c(x.pos, y.pos),
      c(x.pos, y.pos)]), na.rm=TRUE)
    OUT$dSweighted.xy<-((length(x.pos)/(length(c(x.pos, y.pos)))) *
      OUT$dMean.x) + ((length(y.pos)/(length(c(x.pos, y.pos)))) * OUT$dMean.y)
    OUT$Fst.xy<-(OUT$dTotal.xy - OUT$dSweighted.xy) / OUT$dTotal.xy
    OUT$dRelative.xy<-OUT$dMean.xy - OUT$dSweighted.xy
    return(OUT)
}