Hi,

I was trying to do a Significance Analysis of Microarray data for some data (~20,000 genes x 8 total samples). My samples (n=4) belong to either of two groups (A and B). I have two replicates for each sample. So, my samples are:

Sample1_A_replicate1
Sample2_B_replicate1
Sample3_A_replicate1
Sample4_B_replicate1
Sample1_A_replicate2
Sample2_B_replicate2
Sample3_A_replicate2
Sample4_B_replicate2

I want to do a differential gene expression analysis. In the siggenes/SAM documentation, it doesn't mention replicates. How do I go about specifying the class labels?

Should I just use 1,2,1,2,1,2,1,2? Else, how should I specify the class label?

thanks!

Neither siggenes nor samr can correctly handle technical replicates, as far as I know. If you used the classlabel as you describe then you will consider technical replicates as if they were biological replicates, which all things equal should bias you towards the alternative hypothesis. In other words, you will likely get more significant hits than you should, because technical replicates should look more similar, and will shrink your variance estimate.

An alternative is to take the average of the replicates, which is probably a more reasonable thing to do. That will probably reduce the variance as well, but perhaps less than treating technical replicates as if they were biological. Another alternative is to use the limma package, which can accommodate technical replicates.

Thanks James!