On the relationship between perfect matches and mismatches in Affymetrix Genechips.

A relationship which links the fluorescence intensity histograms for perfect match (PM) and mismatch (MM) probes in Affymetrix Genechips is derived using inputs from physical-chemistry as the Langmuir and the Nearest Neighbor models. This relationship is in good agreement with experimental data from few dozens of chips belonging to about 10 different organisms. Principles of physical-chemistry impose some constant value for the average ratios of PM and MM intensities. Experimental data, however, show quite some variations in these parameters, although they follow the same inequalities as expected from hybridization free energies for oligonucleotides melting in solution. It is suggested that the anomalous experiment to experiment differences are due to 1) problems with chip design and 2) excessive fragmentation of the target in solution. The histogram analysis developed may be a useful preprocessing step to evaluate the global quality of the experimental data, prior to the calculation of the gene expression level.