Question

In R I have a data frame with 9 named columns, describing experimental data. The first column contains gene names, and the following 8 columns contain gene expression values. The ultimate goal is to sort the data frame by the standard deviation of the expression values.

Basically, I want to compute the standard deviation of columns 2 to 9 and sort the entire data frame according to these values. How can I do that?

My first idea was to add a tenth column with the standard deviation, computed using the sd() function, then sorting the table, finally deleting the tenth column again. But I can not figure out how to do that.

Answer 1

The key commands are: apply, order, and some row rearrangement.

##Create some dummy data
##You should always try and include some test data in your questions
R> dd = as.data.frame(matrix(rnorm(80), ncol=8))
R> dd = cbind(GENE = LETTERS[1:10], dd)
R> head(dd, 2)
   GENE    V1     V2      V3      V4      V5     V6      V7      V8
 1    A 1.693 1.2977  1.2220  0.4877 -1.7076 1.7796  0.7980 0.08643
 2    B 1.987 0.1545 -0.2173 -0.5959  0.7274 0.2757 -0.5391 0.56054

##Work out the sd for columns 2 to 9 using apply
##Use "order" to reorder the rows
R> dd1 = dd[order(apply(dd[,2:9], 1, sd)),]
##Check the new order
R> apply(dd1[,2:9], 1, sd)
     8      7      5      9      2      1      4      6     10      3 
0.5197 0.7128 0.8149 0.8210 0.8624 0.8808 0.9804 1.2058 1.5086 1.6191 
R> head(dd1, 2)
  GENE      V1      V2     V3      V4      V5      V6       V7       V8
8    H -0.3869  0.6206  0.279 -0.3867 -0.4915 -1.0979 -0.07696 -0.09097
7    G -1.2966 -1.1279 -1.082 -0.4739  0.2717 -0.1365  0.38614  0.38445