Probing the nature of microarray-based oligonucleotides.

Barrett and Kawasaki's recent review [1] of the three forms of probe (short oligonucleotides, long oligonucleotides and cDNAs) used on microarrays raised several points of discussion. None was more important than their assertion that the microarray field is in great need of standardization. Many would point to the nomenclature, annotation and sequence of array probes as being the prime example of this lack of standardization – and I agree. However, I always use the word 'probe' gingerly when referring to DNA arrays, and suggest that when standardization does finally occur we should clarify what to call those dots on the arrays. Historically, scientists labeled a defined oligonucleotide sequence, called it 'the probe' and hybridized it to a target population of nucleic acid. With microarrays, the process is reversed: the target population is labeled, and hybridized to a collection of defined oligonucleotide sequences. This inversion of the original procedures has confused many people regarding the definition of a probe. For example, does 'probe' refer to the defined gene sequence(s) under investigation or does it refer to the labeled component of the hybridization mix? Barrett and Kawasaki, like many others, use the word 'probe' to describe the labeled component. The Henderson's Dictionary of Biological Terms [2] defines 'probe' as a 'well-defined, labeled fragment of DNA or RNA used to find and identify corresponding sequences in nucleic acids by selectively hybridizing with them'. 'Well-defined' seems to be the key here, and suggests to me that the dots on the arrays, and not the labeled population, are the probes. Perhaps this indicates that the definition of a 'probe' in the context of molecular biology needs modernizing and/or redefining. Undoubtedly there are those who think that this point is a disdainful lesson in semantics. Of more interest to them might be the authors' more pragmatic suggestion that the long oligonucleotide-based approach is a good candidate to replace cDNA arrays. This might well be the case for standard laboratory species whose genomic sequences have been characterized extensively. However, cDNA-based arrays might find a more enduring niche among those working on species whose genomes have not been so well characterized. For such species, cDNA-based arrays offer two advantages. (1) Commercially available arrays for human and standard laboratory species might be useful in cross-hybridization experiments. Such hybridizations are more likely to be successful with longer sequences because there is more opportunity for sufficiently similar sequences to exist and …