data-raw/legacy_scripts/correlation_between_gex.R

In Syksy/curatedPCaData: Curated Prostate Cancer Data

####
#
# Pre-calculation of correlation_between_gex.Rmd to avoid overtly long vignette build times
#
####

library(curatedPCaData)

calculate_correlation <- function(dataset, gene) {
  name <- deparse(substitute(dataset))
  data_subset <- as.data.frame(t(dataset[gene, ]))

  data_subset$sample <- rownames(data_subset)
  data_subset[, 1] <- as.numeric(data_subset[, 1])

  dataset <- as.data.frame(t(dataset))


  cor_values <- numeric(0)

  for (i in 1:ncol(dataset)) {
    indvidual_cor_values <- print(cor(dataset[, i], data_subset[, 1], method = "spearman"))
    cor_values <- c(cor_values, indvidual_cor_values)
  }

  cor_matrix <- data.frame(matrix(NA, nrow = ncol(dataset), ncol = 2))
  colnames <- colnames(dataset)[1:ncol(dataset)]
  colnames <- as.vector(colnames)
  cor_matrix[, 1] <- colnames
  cor_matrix[, 2] <- cor_values
  # paste0("spearman_correlation_coeff_",deparse(substitute(dataset)))
  colnames(cor_matrix) <- c("gene", name)
  return(cor_matrix)
}


correlation_of_correlations <- function(gene) {
  tcga <- as.data.frame(curatedPCaData::mae_tcga[["gex.fpkm"]])
  taylor <- as.data.frame(curatedPCaData::mae_taylor[["gex.rma"]])
  abida <- as.data.frame(curatedPCaData::mae_abida[["gex.relz"]])
  ren <- as.data.frame(curatedPCaData::mae_ren[["gex.relz"]])
  barbieri <- as.data.frame(curatedPCaData::mae_barbieri[["gex.relz"]])
  igc <- as.data.frame(curatedPCaData::mae_igc[["gex.rma"]])
  friedrich <- as.data.frame(curatedPCaData::mae_friedrich[["gex.logq"]])
  barwick <- as.data.frame(curatedPCaData::mae_barwick[["gex.logq"]])
  chandran <- as.data.frame(curatedPCaData::mae_chandran[["gex.rma"]])
  icgcca <- as.data.frame(curatedPCaData::mae_icgcca[["gex.rma"]])
  kim <- as.data.frame(curatedPCaData::mae_kim[["gex.rma"]])
  kunderfranco <- as.data.frame(curatedPCaData::mae_kunderfranco[["gex.logr"]])
  sun <- as.data.frame(curatedPCaData::mae_sun[["gex.rma"]])
  true <- as.data.frame(curatedPCaData::mae_true[["gex.logr"]])
  wallace <- as.data.frame(curatedPCaData::mae_wallace[["gex.rma"]])
  wang <- as.data.frame(curatedPCaData::mae_wang[["gex.rma"]])
  weiner <- as.data.frame(curatedPCaData::mae_weiner[["gex.rma"]])

  cor_matrix_tcga <- calculate_correlation(tcga, gene)
  cor_matrix_taylor <- calculate_correlation(taylor, gene)
  cor_matrix_abida <- calculate_correlation(abida, gene)
  cor_matrix_ren <- calculate_correlation(ren, gene)
  cor_matrix_barbieri <- calculate_correlation(barbieri, gene)
  cor_matrix_igc <- calculate_correlation(igc, gene)
  cor_matrix_friedrich <- calculate_correlation(friedrich, gene)
  cor_matrix_barwick <- calculate_correlation(barwick, gene)
  cor_matrix_chandran <- calculate_correlation(chandran, gene)
  cor_matrix_icgcca <- calculate_correlation(icgcca, gene)
  cor_matrix_kim <- calculate_correlation(kim, gene)
  cor_matrix_kunderfranco <- calculate_correlation(kunderfranco, gene)
  cor_matrix_sun <- calculate_correlation(sun, gene)
  cor_matrix_true <- calculate_correlation(true, gene)
  cor_matrix_wallace <- calculate_correlation(wallace, gene)
  cor_matrix_wang <- calculate_correlation(wang, gene)
  cor_matrix_weiner <- calculate_correlation(weiner, gene)



  # put all data frames into list
  df_list <- list(
    cor_matrix_tcga, cor_matrix_taylor, cor_matrix_abida, cor_matrix_ren,
    cor_matrix_barbieri, cor_matrix_igc, cor_matrix_friedrich, cor_matrix_barwick,
    cor_matrix_chandran, cor_matrix_icgcca, cor_matrix_kim,
    cor_matrix_kunderfranco, cor_matrix_sun, cor_matrix_true, cor_matrix_wallace,
    cor_matrix_wang, cor_matrix_weiner
  )

  combined_cor_list <- Reduce(function(x, y) merge(x, y, all = F), df_list)
  rownames(combined_cor_list) <- combined_cor_list[, "gene"]
  combined_cor_list <- combined_cor_list[, -1]

  combined_cor_list[, c(1:17)] <- sapply(combined_cor_list[, c(1:17)], as.numeric)
  return(combined_cor_list)
}

library(corrplot)
combined_cor_list <- correlation_of_correlations("AR")
corr_of_corr <- cor(combined_cor_list)

png("images/correlation_between_gex.png", width = 1000, height = 1000)
corrplot::corrplot(corr_of_corr, method = "color")
dev.off()

save.image("correlation_between_gex.RData")