An Ambiguous View of Protein Architecture

Abstract

Qualifying something as ambiguous means that several interpretations are possible to describe it. We have proposed that such view might be used to analyze protein architecture into structural domains (Postic et al., Science Adv, 2017).The idea of dividing a protein structure into subunits, called domains, was introduced more than four decades ago by Donald B. Wetlaufer, who defined protein domains as structurally compact and separate regions of the macromolecule. However, since this geometrical definition, many manual and automated assignments of structural domains have been proposed, based on criteria such as folding independence, function, thermodynamic stability, or motions. As a result, the same protein may be viewed differently depending on the criteria used for domain partitioning. Therefore, it may be more appropriate to consider that protein structures may be ambiguous and have different, but equally valid, domain delineations.Unfortunately, the a priori, based on our knowledge, can hamper our faculty to recognize more than one way to split an object—a protein, in this respect— thus leading to miss interesting prospects.We have recently proposed a new automated method able to produce alternative domain delineations of protein structures (Postic et al., Science Adv, 2017). Our conceptually new method is based on the hierarchical merging of Protein Units, which describe the protein structure organization at an intermediate level, between secondary structures and domains.We have extensively benchmarked our method using several reference datasets of domain assignments made by either human experts or algorithms. We have shown the significance of our approach with examples covering a wide range of problems related to protein structure, folding, function, and evolution.