[[mkatari-bioinformatics-august-2013|Back to Manny's Bioinformatics Workshop Home]]

====== Volcano Plot using sample Microarray Data ======

In this exercise we will practice our R-coding skills by analyzing a [[https://docs.google.com/file/d/0B172nc4dAaaONlZ4QXByRThPNXM/edit?usp=sharing|sample microarray dataset]] to create a [[http://en.wikipedia.org/wiki/Volcano_plot_(statistics)|Volcano plot]]. A volcano plot is a great visual aid to identify genes that are differentially expressed. The Log Fold Change is plotted on the x-axis and the -1 * Log base 10 of the p-value from a statistical test is plotted on the y-axis. Significantly differentially expressed genes have a p-value less than a cutoff (usually 0.05) and a log fold change of atleast 1.5x (plus or minus).

We will separate the exercise into 5 different steps.
  - Load the file into the R workspace
  - Calculate the log fold change for each gene
  - Calculate the p-value for each gene using t-test
  - Create the Volcano plot
  - Create a boxplot of expression values from genes that are significantly induced.

===== Read the file into R workspace =====

<code>
expvalues = read.table("expvalues.txt", header=T)
</code>

===== Calculate log fold change for each gene =====

==== Using a loop ====

<code>
expvaluesfc = numeric()
for (i in 1:nrow(expvalues)) {
  trtmean = mean(as.numeric(expvalues[i,4:6]))
  ctrlmean=mean(as.numeric(expvalues[i,1:3]))
  expvaluesfc[i] = log2(trtmean/ctrlmean)

}
</code>

===== Calculate p-value using t-test for each gene =====

==== Using a loop ====

<code>
expvaluestt= numeric()

for (j in 1:nrow(expvalues)) {

  t.test(as.numeric(expvalues[j,1:3]),
         as.numeric(expvalues[j,4:6]), 
         var.equal=TRUE)->sample.tt
  
  expvaluestt[j]=sample.tt$p.value
}
</code>

===== Creating a volcano plot =====

<code>
expvalueslog = -1*log(expvaluestt)
plot(expvaluesfc, expvalueslog)
abline(v=-1, col="blue")
abline(v=1, col="blue")
abline(h=29.95, col="red")
</code>

===== A boxplot of induced genes =====

<code>
expvaluefc2tt05 = expvaluesfc > 1 & expvaluestt < 0.05
expvalues[which(expvaluefc2tt05),]->inducedGenes
boxplot(inducedGenes)
</code>