Outliers involving the Poly(A) effect among highly-expressed genes in microarrays

The Poly(A) effect is a cross-hybridization artifact in which poly(T)-containing molecules, which are produced by the reverse transcription of a poly(A)+ RNA mixture, bind promiscuously to the poly(A) stretches of the DNA in microarray spots. It is customary to attempt to block such hybridization by adding poly(A) to the hybridization solution. This note describes an experiment intended to evaluate circumstances under which the blocking procedure may not have been successful.

The experiment involves a spot-by-spot comparison between the hybridization signals obtained by hybridizing a microarray to: (1) end-labeled oligo(dT), versus, (2) cDNA prepared from muscle tissue. We found that the blocking appears to be successful for the vast majority of microarray spots, as evidenced by the weakness of the correlation between signals (1) and (2). However, we found that for microarray spots having oligo(dT) hybridization levels greater than a certain threshold, the blocking might be ineffective or incomplete, as evidenced by an exceptionally strong signal (2) whenever signal (1) is greater than the threshold.

The PolyA effect may be more subtle than simply a hybridization signal that is proportional to the PolyA content of each microarray spot. It may instead be present only in spots that hybridize oligo(dT) greater than some threshold level. The strong signal generated at these "outlier" spots by cDNA probes might be due to the formation of hybridization heteropolymers.