Recurrence Quantification Analysis in structure/function relationships in proteins: an overview of a general methodology applied to the case of beta-lactamase

Protein structure-function relationships have been increasingly scrutinized by a variety of correlational and information theoretic measures. In an effort to extend this methodology, a technique originally developed in non-linear science, recurrence quantification analysis, was combined with traditional principal components analysis to study a large number (56) of TEM-1 beta-lactamase mutants. The hydrophobicity profiles corresponding to the primary structure of 13 naturally occurring mutations partially impairing function, together with 43 artificial non-tolerated mutations were subjected to discriminant analysis, derived from the results of recurrence quantification analysis, coupled to a principal exponents extraction. Eleven (85%) of the naturally occurring mutants and 36 (84%) of the artificial mutants were correctly classified (p < 0.0001). We conclude that this technique may be useful in protein engineering and, in general, in structure-function studies of biopolymers.