distance-source agnostic FunFHMMER
GeMMA Tree cutting into functionally coherent alignments (FunFams).
For the master repo, please see https://github.com/UCL/cath-funfam
See the FunFHMMer Wiki for documentation.
The FunFHMMer algorithm is used to identify functional families in protein domain superfamilies by determining an optimal cut of a hierarchical clustering superfamily tree of sequence relatives by calculating a novel functional coherence index based on conserved positions and specificity-determining positions (SDPs) in sequence alignments [1]. FunFHMMer was used to generate FunFams in the CATH-Gene3D resource (v4.0). FunFams generated by FunFHMMer were shown to generate a more functionally coherent grouping of domain sequences than the other domain classifications. Moreover, it was also shown that the FunFHMMer algorithm is not limited in its use to CATH but can also be used to sub-classify other widely used domain-based classification resources such as Pfam.