While these computational methods offer the advantages of being extremely high-throughput and generally provide accurate broad classifications, exclusive use has led to a significant level of misannotation of enzyme function in protein databases.

[6] Thus although the information now available represents an unprecedented opportunity to understand cellular metabolism across a wide variety of organisms, which includes the ability to identify molecules and/or reactions that may benefit human quality of life, the potential has not been fully actualized.

[8] The EFI is developing an integrated sequence-structure based strategy for functional assignment by predicting the substrate specificities of unknown members of mechanistically diverse enzyme superfamilies.

The computation core performs in silico docking to generate rank-ordered lists of predicted substrates for targeted enzymes using both experimentally determined and/or homology modeled protein structures.

[15] The Haloacid dehydrogenase superfamily contains evolutionarily related enzymes with a Rossmanoid α/β fold with an inserted "cap" region which primarily catalyze metal-assisted nucleophilic catalysis, most frequently resulting in phosphoryl group transfer.

[16] The isoprenoid synthase (I) superfamily contains evolutionarily related enzymes with a mostly all α-helical fold and primarily catalyze trans-prenyl transfer reactions to form elongated or cyclized isoprene products.