Dear list, I have one question regarding microarray data comparisons using Affymatrix yeast whole genome chips. Here are my samples: strains-- WT vs mut, each of them has Total RNA (named as T), translating mRNA (named as P), non-translating mRNA (named as NP) and goes to each chip. Therefore, the final numbers of chips are 2 (strains) X 3 (3 different RNAs) X 2 (repeat) = 12. I have done analyses of WT vs mut with three different RNAs separately. However, my boss wanted me to compare the values of T-P in WT vs mut. * Theoretically*, the outcome should be similar as results from NP comparisons. So, I tricked those normalized dataset in the following steps: 1. import .cel files ( only T and P, no NP) 2. Normalize by RMA 3. convert logged values to unlogged values 4. execute T-P for each strain 5. "shift" values by addition of a fixed values. ( to avoid negative values) 6. log values obtained from step5 7. Scale the dataset (now the values are named as NP2) 8. execute limma to analyze dataset (WT_NP2 vs mut_NP2) (ps. step 3-7 are trying to subtract P signals from Total-RNA signals) Unfortunately, even I do get DEGs, but they are not existing in my previous results. I know this analysis may not make sense in statistic, but at least I want to know if there is any explanation for these data. Any comments or suggestions are really appreciated!!! Hui-Yi [[alternative HTML version deleted]]