Use of hidden correlations in short oligonucleotide array data are insufficient for accurate quantification of nucleic acid targets in complex target mixtures

Nonspecific target binding (i.e., cross-hybridization) is a major challenge for interpreting oligonucleotide microarray results because it is difficult to determine what portion of the signal is due to binding of complementary (specific) targets to a probe versus that due to binding of nonspecific targets. Solving this challenge would be a major accomplishment in microarray research potentially allowing quantification of targets in biological samples. Marcelino et al. recently described a new approach that reportedly solves this challenge by iteratively deconvoluting ‘true’ specific signal from raw signal, and quantifying ribosomal (rRNA) sequences in artificial and natural communities (i.e., “Accurately quantifying low-abundant targets amid similar sequences by revealing hidden correlations in oligonucleotide microarray data”, Proc. Natl. Acad. Sci. 103, 13629–13634). We evaluated their approach using high-density oligonucleotide microarrays and Latin-square designed experiments consisting of 6 and 8 rRNA targets in 16 different artificial mixtures. Our results show that contrary to the claims in the article, the hidden correlations in the microarray data are insufficient for accurate quantification of nucleic acid targets in complex artificial target mixtures.