R/D-matrices.R

In AlphaBeta: Computational inference of epimutation rates and spectra from high-throughput DNA methylation data in plants

#' Constructing D-Matrices
#'
#' Estimating epimutation rates from high-throughput DNA methylation data
#'
#' @param nodelist list of samples, you can find sample file in
#' "extdata" called "nodelist.fn"
#' @param cytosine Type of cytosine (CHH/CHG/CG)
#' @param posteriorMaxFilter Filter value, based on posteriorMax
#' ex: >= 0.95 or 0.99
#' @import dplyr
#' @importFrom  data.table fread
#' @importFrom  data.table fwrite
#' @importFrom  data.table as.data.table
#' @importFrom  stringr str_replace_all
#' @importFrom  gtools mixedorder
#' @importFrom  utils combn
#' @return generating divergence matrices file.
#' @export
#' @examples
#'# Get some toy data
#' file <- system.file("extdata/dm/","nodelist.fn", package="AlphaBeta")
#' df<-read.csv(file)
#' df$filename<-sub("^",paste0(dirname(file),"/"),df$filename )
#' write.csv(df, file = paste0(dirname(file),"tmp_nodelist.fn"),row.names=FALSE,quote=FALSE)
#' file <- system.file("extdata/dm/","tmp_nodelist.fn", package="AlphaBeta")
#' dMatrix(file, "CG", 0.99)


dMatrix <- function(nodelist, cytosine, posteriorMaxFilter) {
    # checking errors
    inputCheck(nodelist, cytosine, posteriorMaxFilter)
    genTable <- fread(nodelist)
    #---------------------------Filter based on meth
    genTable <- genTable %>% filter(genTable$meth=="Y")
    #---------------------------
    pairs <- combn(genTable$filename, 2)
    final_ds <- runMatrix(pairs, cytosine, posteriorMaxFilter, genTable)

    final_ds<-final_ds[mixedorder(final_ds$X1),]
    colnames(final_ds)<-(c("pair.1", "pair.2", "D.value"))

    #dmatrix <- dMsaveResult(final_ds, cytosine, posteriorMaxFilter)
    message("generating d-matrics done.\n")
    rm(genTable)
    return(final_ds)
}


runMatrix <- function(pairs, cytosine, posteriorMaxFilter,genTable){
     flag=TRUE
     name_ds<-NULL
     pair_len <- length(pairs)/2
       for (i in seq_len(pair_len)){
        df <-pairs[,i]
        name_ds[1] <- getNames(df[1], genTable)
        name_ds[2] <- getNames(df[2], genTable)

        cat(paste0("Reading sample: ",name_ds[[1]], " and ",
         name_ds[[2]], " ( ", i , " out of ",length(pairs)/2," pairs )"),"\n")

        cytosine<-as.character(cytosine)
        file_A <-DM.dataRead(df[1],cytosine, posteriorMaxFilter)
        file_B <-DM.dataRead(df[2],cytosine, posteriorMaxFilter)
        # check ad replace pattern in Data-set A
        returnFile <- statusStringCheck(file_A, file_B)
        file_A <- returnFile[[1]]
        file_B <- returnFile[[2]]
        cat("Computing divergence matrix...\n")
        file_A$seqnames<-as.character(file_A$seqnames)
        file_B$seqnames<-as.character(file_B$seqnames)

        tmp_db <- as.data.table(
                  inner_join(file_A, file_B, by = c("seqnames","start","strand")))
        #set status 0=rows is same, 1=M/U 2=I
        rm(file_A,file_B,returnFile)

        tmp_db$state <- ifelse(tmp_db$status.x==tmp_db$status.y,0,
                        (ifelse((tmp_db$status.x=="I" | tmp_db$status.y=="I" ),2,1)))

        # substract number of Intermediate in data-set
        number_none_inter <- sum(tmp_db$state==1)     #M --> U OR U --> M
        number_intermediate <- sum(tmp_db$state==2)   #Intermediate
        Total <- NROW(tmp_db$state)                   #Total rows
        # number of (M --> U or U --> M  and 1/2 I ) / total of rows
        D <- (number_none_inter+(number_intermediate*0.5))/Total
        # reformat
        D <-floorDec(as.numeric(D),5)
          if (flag == TRUE){
             tmp_big<- data.frame(matrix(ncol = 3, nrow = 1))
             tmp_big$X1<-name_ds[[1]]
             tmp_big$X2<-name_ds[[2]]
             tmp_big$X3<-D
             flag = FALSE
          } else {
          tmp<-NULL
          tmp<-list(name_ds[[1]],name_ds[[2]],D)
          tmp_big<-rbind(tmp_big,tmp)
          rm(tmp_db)
          }
         cat(paste0(name_ds[[1]], " and ",name_ds[[2]], " is done! \n"))
         cat("|--------------------------------------------------|\n")
         name_ds<-NULL
       }
    return(tmp_big)
}