Another seemingly straightforward filter has not been used as commonly. Given the accelerating rate of new annotations in DNA databases, it is not uncommon that a probe or probe set used on an older microarray is no longer specific for a single gene. This can happen if the probe was either designed without knowledge of the entire genome or designed against a piece of chromosome instead of an expressed sequence. As an example, probes made against GenBank accession AF053356 might represent the expression of any of 12 expressed products, since this sequence in GenBank represents a piece of chromosome 7. At the time of the microarray construction, however, this region may have been thought to represent only a single gene.
Since without knowledge of the exact probe sequences it is not clear what an expression measurement on a probe set made from a piece of chromosome means, one can argue for removal of these types of probe sets from analysis at the start. However, this requires the use of probe databases early in the analysis, which is not easily performed.
An expression value if the variable whose entropy is being calculated is a measure of gene expression, but any kind of variable, even a noncontinuous or categorical variable such as eye color, can be used in an entropy calculation.
Was this article helpful?