Dynamic Specification of Initial Structures in Parallel Cascade Selection Molecular Dynamics (PaCS-MD) Efficiently Promotes Biologically Relevant Rare Events

Abstract

Abstract Parallel Cascade Selection Molecular Dynamics (PaCS-MD) is an enhanced conformational sampling method for generating structural transition pathways between a given reactant and a product. To promote structural transitions from the reactant to the product, PaCS-MD repeats cycles of conformational resampling of (1) reasonable initial structures and (2) short-time MD simulations restarting the structures with renewed velocities. Appropriately setting a number of these initial structures is essential for efficient PaCS-MD. We propose a novel, optimal algorithm for specifying suitable initial structures, ninitial, to find structural transitions; this is referred to as dn-PaCS-MD. PaCS-MD typically has a fixed ninitial, while in dn-PaCS-MD it is dynamically reset according to a minimum root mean square deviation (RMSD), defined as an RMSD derivative. dn-PaCS-MD, with the dynamic ninitial accelerates structural transitions to the product, compared to the original PaCS-MD with the fixed ninitial, as confirmed in Chignolin protein-folding. This algorithm was also applied to Human-β-defensin 2, whose X-ray and NMR structures differ in the N-terminal region. Finally, we compared forward and backward transition pathways from the former to the latter generated by our method, and estimated a free energy barrier between them, creating new possibilities to visualize the NMR and X-ray structures in solution.