Exon signal enhancers (ESEs) and exon splicing silencers (ESSs) are known to guide the process of exon selection during the mRNA precursor splicing. A number of these elements are active from human to yeast, suggesting that this mechanism evolved a long time ago, which makes it a good target for phylogenetic footprinting. At the same time phylogenetic footprinting can be challenging as the conservation in the exons can be very high simply because of the protein conservation.

Here we describe an algorithm that applies several filters to overcome the problem. First, we select only those exons, which are sufficiently conserved in several vertebrae organisms. We assume it is likely those exons will be constitutively expressed, therefore there are more chances they will contain ESEs. Next we define the exons that are not under evolutionary pressure under in any directions in an attempt to evade sequences that are possibly strongly biased based on the protein composition.

Finally, based on the selected exons we construct a bipartite weighted graph using the hamming distances between every window in every exon. To further minimize the effect of the protein conservation we apply weight to each position in the coding triplet. Finally we find the cliques within the graph and use them to construct minigenes, which in the example we show confirm the presence of an ESE.