Should I get the mean of 2 log2 transformed RPKM values?

Question

[BioinfGuru]

I want to calculate tissue specificity and have 2 biological replicates for each gene in each tissue. I first removed any genes that have a rpkm below 1.0 and then log2 transformed all remaining rpkm values. My questions are: is it appropriate to get the average (mean) of the 2 log2rpkm values for each gene. Would I be better averaging the raw rpkm values? Should I be including SD or variance?

Thank you. Kenneth

[Satyajeet Khare]

Hi Kenneth, It probably should not matter but I guess standard practice is to average RPKM and then log transform. With only two biological replicates, SD may not mean much. Except, including statistics can reduce the number of hits you will have to test in lab.