hi all, sometime ago someone asked about thoughts related to the use of mismatch probes (MMs) in affy arrays. The developmental package gcrma provides an implementation of a version of RMA that uses MMs. It also uses sequence information. the new method appears to improve the performance of RMA in terms of accuracy (estimated fold changes are bigger). the method is described here: http://www.biostat.jhsph.edu/~ririzarr/papers/gcpaper.pdf the code (which is very slow) is still under development thus feedback is greatly appreciated. cheers, rafael