I am doing some statistical analysis of RNASeq data and I have run into a bit of a problem. Instead of looking at comparison of individual genes I want to compare groups of genes. The setup is as follows: I have two independent groups of genes. For each gene I have three measures from three biological replicates. I want to compare expression between groups. Sample table with design: https://ibb.co/myCssd
Right now I'm averaging biological replicates and performing Mann-Whitney U-test, but I was wondering if there is a better solution? The other problem is that I can not really compare absolute expression values like FPKM, so I'm looking at % maximum expression instead. This works nicely on averages, but would not really work on individual replicates due to outliers...
Any suggestions would be greatly appreciated.
A bit of background: I have 5 time points, 3 replicates each. I know that between time point 2 and 3 there is a transcription factor which is expressed. I want to see the effect it has on the target genes, compared to the control genes at time point 3. I essentially want to see if there is a difference in expression between target (bound by transcription factor) and control (not bound by the transcription factor) genes at time point 3. At this point I calculate the mean of three replicates for all the genes. Split the genes into target and control sets and then use Mann-Whitney U test to see if there is a difference in underlying distributions. Also, because the absolute values are difficult to compare I scale them to % of maximum expression. So, I’m essentially comparing the difference in distributions of % of maximum expression at point 3. One of my problems is that instead of incorporating the replicates I’m just using the mean. I was wondering if there is a cleaner way performing analysis like that?