Keck-UNM Genomics Resource

Affymetrix All-Exon Arrays

The new all-exon microarrays from Affymetrix contain probesets for 1.4 million well characterized and predicted exons in the human, mouse and rat genomes. The probesets are organized into transcript clusters, which represent the detectable RNAs encoded by the genomes.

The all-exon arrays provide users with the opportunity to meaure both gene expression and alternative RNA splicing, opening up an entirely new way to analyze changes in RNA levels.

What are the differences between conventional 3'-arrays and the new all-exon arrays?

The conventional Affymetrix GeneChips (e.g. U133 Plus 2.0 arrays) have 11 or more perfect match and an equal number of mismatch probes for each transcript, usually located near the 3'-end of the mRNA, most often in the 3'-untranslated region or in the last exon. By weighting the detection to the 3'-end, Affymetrix made sure that the labeling would work well, even if the RNA quality was compromised or if the reverse transcription reaction was inefficient. Having the mismatched probes allowed the background from each probe to be subtracted. Having so many probes allowed statistical tests to determine whether each mRNA was significantly detectable above background.

In contrast, the new all-exon arrays (e.g. Human Exon 1.0 ST) have 4 probes each for 1.4 million well characterized and predicted exons in the human genome, and similar sets of probes for the mouse and rat GeneChips. Instead of mismatched probes, the new arrays rely on using collections of other probes with similar G+C content to determine background levels. These comparisons are used for a "Detectable Above BackGround" (DABG) calculation which replaces the old Present/Marginal/Absent call. In addition, the probesets representing all the possible exons are organized through the use of "metaprobesets" into transcript clusters with increasing levels of complexity. The metaprobesets are essentially just lists of transcript clusters and the probesets that are associated with them, and they will evolve and change as the genome annotation becomes more complete. Amongst the metaprobesets, the most basic or "core" set includes only those exons that are known to be included within well-characterized transcripts. The next most complex "extended" set also includes exons that are part of alternatively spliced mRNAs or that are predicted to be included in alternative transcripts. Finally, the "full" set contains all the exons that can be associated with transcript clusters, even if there is little evidence that they are biologically relevant. So, in the new all-exon terminology, probesets are exons and transcript clusters are genes or mRNAs.

What are the new all-exon arrays used for?

The new Affymetrix all-exon arrays (e.g. Human Exon 1.0 ST) can be used to measure both transcript-level and exon-level changes in gene expression. Unlike the conventional 3'-arrays, transcript-level changes in expression are determined by calculating the mean change in all the exons in a transcript cluster. Thus, if a gene has 15 exons or probesets in its transcript cluster, the mean of all 15 measurements, each of which is in turn a mean of 4 independent measurements, is used to calculate the expression level of the overall transcript. In this hypothetical example, 60 different measurements, spread throughout the length of the mRNA, would be used to calculate the changes in expression of this single transcript.

In addition to transcript-level changes, the new all-exon arrays can measure changes in alternative splicing, by following the expression of each exon, or probeset, independently.

Using the all-exon arrays in slightly (about 25%) more expensive than using the older 3'-end arrays, but the data is much more extensive and many researchers will likely choose to use the all-exon arrays for all their Affymetrix GeneChip experiments.

• Affymetrix GeneChip Human Exon 1.0 ST Array
• Exon Array Data Analysis
• Integrated Genome Browser (IGB)
• GeneChip Compatible Software