Exon Array Background Correction
Revision Date: 2005-09-27
Revision Version: 1.0
Affymetrix GeneChip® Exon Array Whitepaper Collection:
1 of 5
Exon Array Background Correction
We expect to obtain similar performance in detection of differential changes for
most applications on exon arrays using the background probe collection (BGP)
as opposed to specific mismatch probes (MM). (For example, the GeneChip®
Human Exon 1.0 ST Array has a “genomic” and an “antigenomic” background
probe collection. See Appendix for more information on the background probe
collections.)
In this whitepaper BGP probes are used to perform a PM-GCBG correction. PM-
GCBG refers to the use of the median BGP intensity for BGP probes with the
same GC content as the perfect match (PM) probe.
MM probes provide controls similar in sequence and location which are useful in
removing bias; this takes one probe per informative probe. Because the MM
probes hybridize to the perfect match (PM) target, when MM intensities are
subtracted from PM intensities the PM response to true target slightly reduced.
BGP probes provide a rough estimate of background based on coarsely modeled
sequence, without taking into account any spatial variation. Use of BGP over
MM probes requires roughly 50% less space on the array; thus use of BGP
probes allows roughly a 2 fold increase in PM content.
Background estimates from BGP and MM probes may not be accurate for
specific PM probes leading to biases in the background estimates results. We
therefore look at the typical distribution of error in modeling background, as well
as the downstream results on the ability to detect differential change. These
results are based on a research array design using a pre-optimized version of the
whole transcript assay.
We examine the performance of these two methods with a controlled experiment
on a research chip using spiked-in transcripts. This experiment consists of 9 full-
length clones which are spiked in at several different levels of relative
abun