R/base_compareData.R

In huangwb8/GSClassifier: Gene Signature Classifier

#' @title Compare data from the same source but different scale
#' @description Compare data from the same source but different scale
#' @param X1 Matrix with sample col and gene row
#' @param X2 Matrix with sample col and gene row
#' @param names Label of X1 and X2
#' @param width,height Plot parameters
#' @param savePath Plot saved path
#' @importFrom plyr alply
#' @importFrom ggplot2 ggplot geom_point aes stat_smooth labs theme_bw theme
#'   element_text element_rect
#' @author Weibin Huang<\email{654751191@@qq.com}>
#' @examples
#' ## PAD genes
#' PAD <- readRDS(system.file("extdata", 'PAD.rds', package = "GSClassifier"))
#' tgene <- as.character(unlist(PAD$geneSet))
#' X1 <- stad_fpkm[tgene,]
#' X2 <- stad_tpm[tgene,]
#'
#' ## Compare FPKM with TPM matrix
#' res <- compareData(X1,X2,
#'                    names=c('FPKM','TPM'),
#'                    width = 10,
#'                    height = 10,
#'                    savePath = '.')
#' res2 <- compareData(log2(X1+1),log2(X2+1),
#'                     names=c('log2FPKM','log2TPM'),
#'                     width = 10,
#'                     height = 10,
#'                     savePath = '.')
#' @export
compareData <- function(X1,
                        X2,
                        names = c('X1', 'X2'),
                        width = 10,
                        height = 10,
                        savePath = '.') {
  ## Data
  coGene <- intersect(rownames(X1), rownames(X2))
  coSample <- intersect(colnames(X1), colnames(X2))
  X1 <- X1[coGene, coSample]
  X2 <- X2[coGene, coSample]

  ## Correlation of expression for every gene
  compareOneGene <- function(X1, X2, gene) {
    # gene='ENSG00000136167'
    x1 <- as.numeric(X1[gene, ])
    x2 <- as.numeric(X2[gene, ])
    res <- cor.test(
      x1,
      x2,
      alternative = "two.sided",
      method = "spearman",
      conf.level = 0.95
    )
    # res2 <- data.frame(
    #   Cor = res$estimate,
    #   P.value = res$p.value
    # )
    size = 30
    p <- ggplot(data.frame(x1 = x1, x2 = x2), aes(x = x1, y = x2)) +
      geom_point(alpha = 0.5,
                 size = 11,
                 colour = "#FB8072") +
      stat_smooth(formula = y ~ x, method = 'glm') +
      labs(x = names[1],
           y = names[2],
           title = gene) +
      theme_bw() +
      theme(
        plot.title = element_text(
          size = size,
          colour = "black",
          face = "bold",
          hjust = 0.5
        ),
        axis.text = element_text(
          size = size / 15 * 12,
          colour = "black",
          face = "bold"
        ),
        axis.title.x = element_text(
          size = size,
          colour = "black",
          face = "bold"
        ),
        axis.title.y = element_text(
          size = size,
          colour = "black",
          face = "bold"
        ),
        legend.text = element_text(
          size = size / 15 * 12,
          colour = "black",
          face = "bold"
        ),
        legend.title = element_text(
          size = size / 15 * 12,
          colour = "black",
          face = "bold"
        ),
        legend.position = 'right',
        strip.background = element_rect(fill = "white"),
        strip.text = element_text(
          size = size / 15 * 12,
          colour = "black",
          face = "bold"
        )
      )

    # Output
    l <- list(Data = res,
              Plot = p)
    return(l)

  }
  l <-
    alply(as.matrix(coGene), 1, function(x)
      compareOneGene(X1, X2, x))
  names(l) <- coGene

  ## Merge plot
  cairo_pdf(
    paste0(
      savePath,
      '/Plot of compareData_',
      paste0(names, collapse = '-'),
      '.pdf'
    ),
    width = width,
    height = height,
    onefile = T
  )
  for (i in 1:length(l))
    print(l[[i]]$Plot)
  dev.off()

  ## Output
  return(l)
}