SAMPL6 host–guest blind predictions using a non equilibrium alchemical approach

Abstract

In this paper, we compute, by means of a non equilibrium alchemical technique, called fast switching double annihilation methods (FSDAM), the absolute standard dissociation free energies of the the octa acids host-guest systems in the SAMPL6 challenge initiative. FSDAM is based on the production of canonical configurations of the bound and unbound states via enhanced sampling and on the subsequent generation of hundreds of fast non-equilibrium ligand annihilation trajectories. The annihilation free energies of the ligand when bound to the receptor and in bulk solvent are obtained from the collection of work values using an estimate based on the Crooks theorem for driven non equilibrium processes. The FSDAM blind prediction, relying on the normality assumption for the annihilation work distributions, ranked fairly well among the submitted blind predictions that were not adjusted with a linear corrections obtained from retrospective data on similar host guest systems. Improved results for FSDAM can be obtained by post-processing the work data assuming mixtures of normal components.

In-gap state generated by La-on-Sr substitutional defects within the bulk of SrTiO3

“…These good performances of the NEW approach have been confirmed in the second round of the SAMPL6 challenge for the LogP. 21 In spite of the fact that the rare applications of NEW methods have been quite successfully for solvation and binding free energies of complex molecular systems, [32][33][34][35][36][37][38][39] the vast majority of practitioners still rely upon FEP-based technology, in most cases systematically and incautiously underrating the sampling issue. 26 This is probably due to incomplete or far-stretched analysis studies 31,40 where NEW methods have been shown, in specific examples, to be computationally comparable to or less efficient than equilibrium approaches.…”

“…NEW approaches are still rarely used in free energy calculations. In the recent SAMPL6 initiative for blind prediction of host-guest binding free energies 25 , the only NEW submission 39 turned out be one of the best performing methods in the challenge. These good performances of the NEW approach have been confirmed in the second round of the SAMPL6 challenge for the LogP.…”

“…In spite of the fact that NEW methods have been already applied quite successfully to solvation free energies 33,34 as well to drug-receptor [35][36][37] , host-guest 38,39 and even protein-protein 32 binding free energies, the vast majority of practitioners still stick to FEP-based technology. In the recent LogP/SAMPL6 challenge, NEW was used only in my submissions.…”

SAMPL7 TrimerTrip Host-Guest Binding Poses and Binding Affinities from Spherical-Coordinates-Biased Simulations

“…The statistical assessment of the modeling (SAMPL) challenges feature the assessments of the sampling and Hamiltonian issues in the computational modeling of solvation free energies, pKa, host-guest systems, partition coefficients, and protein-ligand binding. [90][91][92][93][94][95] The host-guest systems are analogues of protein-ligand complexes. They are smaller and simpler than proteins and ligands.…”

“…92,95 Nonequilibrium free energy simulations in the alchemical space were performed to calculate the host-guest binding affinities. 94 Although the free energy methods are accurate, the mean deviations from the experimental reference are often 2 kcal/mol, 93 which in principle arise from the Hamiltonian issue.…”