Overview of the SAMPL6 host-guest binding affinity prediction challenge

Abstract

The ability to accurately predict the binding affinities of small organic molecules to biological 15 macromolecules would greatly accelerate drug discovery by reducing the number of compounds that must 16 be synthesized to realize desired potency and selectivity goals. Unfortunately, the process of assessing the 17 accuracy of current quantitative physical and empirical modeling approaches to affinity prediction against 18 binding data to biological macromolecules is frustrated by several challenges, such as slow conformational 19 dynamics, multiple titratable groups, and the lack of high-quality blinded datasets. Over the last several 20 SAMPL blind challenge exercises, host-guest systems have emerged as a practical and effective way to 21 circumvent these challenges in assessing the predictive performance of current-generation quantitative 22 modeling tools, while still providing systems capable of possessing tight binding affinities. Here, we present 23 an overview of the SAMPL6 host-guest binding affinity prediction challenge, which featured three supramolec-24 ular hosts: octa-acid (OA), the closely related tetra-endo-methyl-octa-acid (TEMOA), and cucurbit[8]uril (CB8), 25 along with 21 small organic guest molecules. A total of 119 entries were received from 10 participating 26 groups employing a variety of methods that spanned electronic structure and movable type calculations 27 in implicit solvent to alchemical and potential of mean force strategies using empirical force fields and 28 explicit solvent models. While empirical models tended to obtain better performance, it was not possible 29 to identify a single approach consistently providing superior predictions across all host-guest systems and 30 statistical metrics, and the accuracy of the methodologies generally displayed a substantial dependence on 31 the systems considered, arguing for the importance of considering a diverse set of hosts in blind evaluations. 32 Several entries exploited previous experimental measurements of similar host-guest systems in an effort 33 to improve their physical-based predictions via some manner of rudimentary machine learning; while this 34 strategy succeeded in reducing systematic errors, it was not able to generated a corresponding improvement 35 of correlation statistics. Comparison to previous rounds of the host-guest binding free energy challenge 36 highlights an overall improvement in the correlation obtained by the affinity predictions for OA and TEMOA 37 systems, but a surprising lack of improvement in root mean square error over the past several challenge 38 rounds. The data suggests that further refinement of force field parameters and improved treatment of 39 chemical effects (e.g., buffer salt conditions, protonation states) may be required to continue to enhance 40 predictive accuracy. 41 42 1 of 35 Preprint ahead of submission -July 18, 2018 48 Assessment of how much of this inaccuracy can be attributed to fundamental limitations of the force 49 field in accurately modeling energetics ...

Prediction of CB[8] host–guest binding free energies in SAMPL6 using the double-decoupling method

“…The x-axis denotes receipt ID of all submissions. Further statistical analysis can be found in an overview paper for the SAMPL6 host-guest binding affinity prediction challenge in the same issue [11]. …”

“…Since SAMPL3 in 2011, the scope of the SAMPL challenge has expanded from calculating solvation free energies (SAMPL1) [6] and tautomeric states of drug-like small molecules (SAMPL2) [7] to predicting binding free energies of host-guest systems (SAMPL3–6) [8–11]. …”

Rationalizing the “Diels-Alderase” Activity of Pd2L4 Self-Assembled Metallocages: Enabling the Efficient Prediction of Catalytic Scaffolds

“…66 It is important to note that while reasonably predictions of absolute binding free energies have been presented for organic host-guest complexes, errors of at least 1-4 kcal mol −1 have been reported for such systems, mostly due to vibrational anharmonicity and solvent effects. [67][68][69] The binding energy of a general quinone (qn) was defined as ∆Ebind(qn⊂C-X) = E(qn⊂C-X) -…”

Blinded predictions of standard binding free energies: lessons learned from the SAMPL6 challenge

“…The 6th SAMPL (SAMPL6) competition was launched in September 2017. Our group focused on the host-guest leg of this contest, which requested predictions of standard free energies of binding for 27 guests across three different hosts [31]. The host molecules consisted in two octa-acids, OA and TEMOA molecules [32][33][34][35], and a cucurbituril ring clip CB8 [36][37][38][39], as shown in Fig.…”