It is analagous to, and based on, the BioPerl Bio::SeqFeature::SimilarityPair.

It is different from HomologyFeature in that it expresses a relationship between only two sequence regions (rather than >= 2), with one clearly defined as the query sequence and the other as the subject (database hit). These are identified by constant labels in the Alignment. HomologyFeature identifies the related sequence regions by means of an Homology instance which contains an Alignment which uses the HomologyFeatures themselves as labels.

In cases where there is no alignment available, for example when MSPCrunch output or GFF have been used, the EmptyPairwiseAlignment in the EMPTY_PAIRWISE field may be used. This may also be useful if an implementation elides the alignment data for some reason.

Original BioJava version by Keith James. Port to C# by Doug Swisher.