The Rough Guide to In Silico Function Prediction, or How To Use Sequence and Structure Information To Predict Protein Function
Figure 1
Homology based annotation transfer: Problems.
(A) Paralogy problem: Paralogs are more likely to diverge functionally with respect to orthologs. If our putative template is a paralog, the probability that the query has similar function decreases. (B) Moonlighting problem: If the template performs multiple functions, the query could have retained only some of them (and vice-versa, if the query were a moonlighting protein, using a non-moonlighting template would result in an incomplete annotation of the query). (C) Multi-domain proteins problem: If the template is annotated based on the function of a domain that is not aligned to the query, annotation transfer is not possible. (D) Database mis-annotations problem: Database entries may have been mis-annotated; the risk is especially high if annotation was performed automatically via homology transfer.