Comparison of two predictions. From the comparison of two predictions (a) we obtain three differentiated sets of intron assemblies (IAs): the set of IAs that are identical in both transcripts ('A and B'), and two set of the IAs that are in one prediction but not in the other ('A not B' and 'B not A'). When two sets have the same intron with different outside boundaries for the flanking exons these boundaries are taken from the intersection of the exons. Ensembl predictions (b) have in general more than one transcript per gene (two top yellow tracks). The intersecting intron assemblies (IAs) are therefore defined as the longest non-redundant IAs common between the transcripts from either prediction. For the novel IAs we take the longest non-redundant IAs in one that are not present in the other set.
Eyras et al. BMC Bioinformatics 2005 6:131 doi:10.1186/1471-2105-6-131