rExample.R

main = function() {
  # Import the data as a dataframe
  drugData = read.delim(file="gene_expression.txt");
  # Name the rows using the names from the first column of the drugData dataframe
  dimnames(drugData)[[1]] = drugData[,1];
  # Drop the first column of the drugData dataframe because we no longer need it because we have named our rows
  drugData = drugData[,-1]
  # Print some of the drugData for a sanity check
  print(drugData[1:5, 1:4]);
  # Check the class of the drugData to confirm that it is a dataframe
  print(class(drugData));

  # Make the drugData numeric
  drugDataAsVector = as.vector(unlist(drugData), mode="numeric");

  # Create a historgram of all the drugData
  png("histogram.png")
  hist(drugDataAsVector, xlab="gene expression");

  # Create a histogram of the atf1Data
  atf1Data = drugData$ATF1;
  print(atf1Data);
  png("histogramatf1.png")
  hist(atf1Data, xlab="ATF1");

  # Calculate the max, min, and mean of the drugData
  print(max(drugDataAsVector));
  print(min(drugDataAsVector));
  print(mean(drugDataAsVector));

  # Create a boxplot of the first 3 columns and all the rows in drugData
  png("boxplot.png");
  boxplot(drugData[,1:3]);

  # Find the correlation between 2 columns
  cor(drugData[,1], drugData[,2]);
  png("scatterplot.png");
  plot(drugData[,1], drugData[,2]);
  coreraltionMatrix = cor(drugData);
  print(class(coreraltionMatrix));

  # Find the max and min correlation
  maxCor = -1;
  minCor = 1;
  for (geneA in 1:ncol(coreraltionMatrix)) {
    for (geneB in 1:ncol(coreraltionMatrix)) {
      if (geneA != geneB) {
        currentCor = cor(drugData[,geneA], drugData[,geneB])
        if (currentCor > maxCor) {
          maxCor = currentCor;
          print(maxCor)
        }
        if (currentCor < minCor) {
          minCor = currentCor;
          print(minCor)
        }
      }
    }
  }
  print(paste("The max cor is: ", maxCor));
  print(paste("The min cor is: ", minCor));

}

main();