Correlating molecular structures and self-assembly mechanisms via temporal analysis of multidimensional chemical interaction space: application to assemblies of isomeric peptides

Abstract

Computational modelling of self-assembly mechanisms is a promising way to establish chemically meaningful relationships between molecular structures of the building blocks and the final outcomes of the spontaneous assemblies. However, such connections are not immediately apparent, due to the presence of complex interplay involving a multitude of intermolecular interactions with complicated temporal variations. In this paper, we propose a method called temporal analysis of multidimensional chemical interaction space (TAMCIS), which looks at important combinations of interactions, rather than analysing them one at a time. Each molecule was assigned a vector order parameter, with components representing appropriately chosen chemical interactions. The aggregate data were processed using density-based clustering, resulting in “interaction clusters”. Time dependent partitioning of the molecules among these clusters revealed the mechanism in terms of interactions, thereby making a direct connection to the molecular structures of the building blocks. We applied the method to a comparative study of assembly mechanisms of two isomeric hydrophobic tri-peptides in water, namely tri-L-leucine (LLL) and tri-L-isoleucine(III). Initially, both systems started to aggregate via non-bonded interactions through sidechains. But at later stages, they diverged in the interaction space when hydrogen bonding and electrostatic contacts became important. Overall, a stark difference was observed. LLL assembly grew by a combination of interactions. In contrast, the III system primarily utilized one type of hydrogen bonding, leading to β-sheet-like arrangements found in proteins. The TAMCIS provided a clear path for deciphering the origins of emergent complexities in spontaneous self-assemblies from dynamical simulation data.