Classifying substrate specificities of membrane transporters from Arabidopsis thaliana

Membrane transporters catalyze the active transport of molecules across biological barriers such as lipid bilayer membranes. Currently, the experimental annotation of which proteins transport which substrates is far from complete and will likely remain so for much longer. Therefore, it is highly desirable to develop computational methods that may aid in the substrate annotation of putative membrane transport proteins. Here, we measured the similarity of membrane transporters from Arabidopsis thaliana by their amino acid composition, higher sequence order information, amino acid characteristics, or sequence conservation. We considered the substrate classes amino acids, oligopeptides, phosphates, and hexoses. Substrate classification based on the amino acid frequency yielded an accuracy of 75% or higher. Integrating additional information improved the prediction performance to 90% and higher.