Supplemental Results and Discussion

For genome-wide expression studies data handling is important since global effects in the same direction on many genes can skew the results. The standard method recommended by Affymetrix is MAS5. We have not used this method but rather RMA which normalize the data on probe level and not on gene level and will accordingly in theory be less prone to normalize away global effects. It is important to note that if we assume global effects acting on all genes, i.e., a haploid region would cause an increase in expression of all other genes, this effect cannot be measured since we start with a given amount of material (RNA) for labelling. The difference between the haploid region and the rest of the genome can be measured, but whether the remaining genome is globally altered remains unknown. Such global effects could only be measured with one other fixed reference, such as number of cells or amount of DNA and would require that all additional steps, such as labelling, hybridisation and signal read-out are perfectly quantitative. With current methodology global effect acting on all genes cannot be correctly determined.