Why does Roche NimbleGen use Tukeys bi-weight function for scaling ChIP-chip data?
RMA looks at all the arrays in a set and normalizes the data for all the arrays. ChIP-chip is a two color array with the reference on the array and is therefore thought of as a stand alone experiment. RMA is not an appropriate analysis for ChIP chip data. The Tukey bi-weight function is used to account for differences in the dyes on the array, whereas RMA is used to account for differences between arrays so that the arrays can be compared.