Quantum chemical prediction for complex organic molecules

Abstract

Numerous types of quantum chemical calculations and protocols have been successfully applied to computing pK a of small, uncomplicated organic molecules. Here, we argue for the need to shift attention to more challenging molecules that are marked by an interplay of complicating factors such as conformational, tautomeric, steric, and other effects. The challenge is not in choosing the right quantum chemical method and solvation model but in combining the existing methods to simultaneously and accurately describe the breadth of chemical and physical phenomena that give rise to the experimentally observed pK a . The complexity of the phenomena that must be considered begs for the need for a greater automation of pK a prediction workflows. We review our experience with these challenges and outline paths for future progress in the direction of tackling pK a prediction of complex organic molecules.complex, pK a , prediction, protonation, quantum | SI M P L E A N D COM P LE X M OL ECU L E SThere is little doubt that methods of pK a prediction based on quantum chemistry and normally employing density functional theory (DFT) energy calculations have largely achieved their goals when it comes to small, simple organic molecules. Such molecules tend to have fewer than 20-30 atoms, are characterized by a single and well-known functional group responsible for the measured pK a , and are relatively rigid. Their predicted pK a values often lie within 0.5 units of the experimental values, the corresponding protocols are uncomplicated and the interpretation of the results straightforward. [1][2][3] To predict the pK a of these compounds is usually not a goal in itself. Their dissociation constants are known from experiment, and in any case can be accurately and instantly predicted by empirical methods. So these small compounds are used mostly in proof-of-principle studies or for method development.The situation with complex organic molecules is very different. Proteins aside, being a separate topic, complex organic molecules, as far as pK a prediction goes, are characterized by one or more of the following features: 50 atoms and more, considerable flexibility with the possibility of forming intramolecular hydrogen bond and other noncovalent interactions, steric effects, several functional groups that might take part in protonation/ deprotonation equilibria (PDE), symmetry-equivalent functional groups, multiple possible tautomeric forms, including zwitterionic species (which might be hard to describe with implicit solvation models), unusual or novel functional groups or heterocycles, and more than one stereocenter that can influence pK a (see Figure 1). An often quoted figure of 1.4 kcal/mol on the energy scale, [4] (which is a small error for DFT energetics), corresponds to 1 unit on the logarithmic scale (which is quite a large error for pK a prediction), so the phenomena aforementioned must either all be described very accurately or a favorable cancelation of errors must be achieved. In view of the multitude of the complicating ...

Enhancing Water Sampling in Free Energy Calculations with Grand Canonical Monte Carlo

“…42 After using Epik, uncertainty remained for ligands 13 and 21 which was resolved with Schrödinger's macro-pKa prediction protocol. 43 As 2HOG is missing residues 44 to 50 as well as the side chain for TYR 20, PDB entry 2E9V (resolved at 1.9 Å, 0.65 Å backbone RMSD from 2HOG) was used as the starting protein structure for FEP. The oxygen positions of the unperturbed three-water network were also taken from 2E9V.…”

“…Schrödinger's method to determine macro-pKas was run on the remaining ligands to estimate the pKas of the aromatic nitrogens. 43 These predictions, shown in Table 4, have an expected error of about 1 pKa unit, making the protonation state of ligands 2d, 6d, and 8d in solvent uncertain. A correction, described previously, 51 was applied to predicted relative free energies to account for the pKa of the ligands in solvent.…”

Aqueous pKa prediction for tautomerizable compounds using equilibrium bond lengths

“…Table 2 shows the 7-fold CV and external validation statistics for optimal models. These were derived using PLS (4…”

“…Recently 2 , Connolly suggested that a lack of experimental information on both relative tautomer stability and the properties of distinct tautomeric forms are the likely causes of such issues. Tautomeric species present a challenge, not just to empirical-based approaches, but also to those that attempt to solve the pK a prediction problem using first-principles [1][2][3][4][5] . For tools implementing the latter approach (e.g.…”

An explicit-solvent hybrid QM and MM approach for predicting pKa of small molecules in SAMPL6 challenge

“…However, if they are close, the macroscopic pKa is shifted as multiple microscopic transitions contribute to the macroscopic value. Several studies [40, 41] discuss this in greater detail. In our method, we calculate microscopic pKa value for each acid-base pair of microscopic states.…”