Filtering out data that are not reliably detected by setting a threshold for fraction of arrays (in at least one treatment group) in which the probe set is called Present by the MAS5 algorithm is a simple, easy to implement approach that works well to reduce false positives with little cost in loss of true positives. It is superior to using the average signal or RMA value because it is more likely to preserve probe sets turned on or off and it removes probe sets that show cross-hybridization. Another advantage is that filtering by fraction Present is much easier to implement, because general guidelines can be set based upon the experiment size instead of having to examine the distribution of expression values.
