hide
Free keywords:
-
Abstract:
Proteins are usually described and classified according to amino acid sequence, structure or function. Here, we develop a
minimally biased scheme to compare and classify proteins according to their internal mobility patterns. This approach is
based on the notion that proteins not only fold into recurring structural motifs but might also be carrying out only a limited
set of recurring mobility motifs. The complete set of these patterns, which we tentatively call the dynasome, spans a multi-
dimensional space with axes, the dynasome descriptors, characterizing different aspects of protein dynamics. The unique
dynamic fingerprint of each protein is represented as a vector in the dynasome space. The difference between any two
vectors, consequently, gives a reliable measure of the difference between the corresponding protein dynamics. We
characterize the properties of the dynasome by comparing the dynamics fingerprints obtained from molecular dynamics
simulations of 112 proteins but our approach is, in principle, not restricted to any specific source of data of protein
dynamics. We conclude that: 1. the dynasome consists of a continuum of proteins, rather than well separated classes. 2. For
the majority of proteins we observe strong correlations between structure and dynamics. 3. Proteins with similar function
carry out similar dynamics, which suggests a new method to improve protein function annotation based on protein
dynamics
