Fig. 1

Schema of the methodology. A representative sequence is taken from each of the 121 non-enzymatic families previously used by Chakraborty et al. [10]. Its homologs are retrieved from the current version of Uniprot (top) and yearly subsets of these are constructed based on their publication date. The sequences of these “historical” sets of homologs are retrieved and aligned in an attempt to generate the multiple sequence alignments one would have obtained at a given time point in the past. Finally, the methods for detecting SDPs and functional subfamilies are applied to these historical alignments and the results contrasted with structural information on binding sites if available (bottom)