SAMPL7 blind challenge: quantum–mechanical prediction of partition coefficients and acid dissociation constants for small drug-like molecules

Fındık, Basak Koca; Haşlak, Zeynep Pınar; Arslan, Evrim; Avi̇yente, Vi̇ktorya

doi:10.1007/s10822-021-00402-9

Cited by 11 publications

(8 citation statements)

References 65 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Thermodynamic and kinetic information could be extracted via statistical analyses. Biologically relevant processes generally happen at time scales of μs, ms, and even longer, which are inaccessible in modern computational modeling. − Enhanced sampling techniques are often employed to deal with the time-scale problem. ,− Modifications of the Hamiltonian of the simulated system are added to accelerate the phase space exploration, and post-simulation analyses are used to recover the statistics in the original unperturbed ensemble. − The Hamiltonian or model employed to describe the system under investigation determines the accuracy of the simulation outcome. − Although detailed descriptions such as quantum mechanics (QM) calculations are accurate, their practical use is quite limited due to high computational costs. − To obtain sufficient data for post-processing analyses, the approximated mean-field molecular mechanics (MM) force fields are widely employed in molecular dynamics (MD) simulations of biomolecular systems. − …”

Section: Introductionmentioning

confidence: 99%

A General Picture of Cucurbit[8]uril Host–Guest Binding

Sun

Huai²,

He³

et al. 2021

J. Chem. Inf. Model.

View full text Add to dashboard Cite

Describing, understanding, and designing complex interaction networks within macromolecular systems remain challenging in modern chemical research. Host− guest systems, despite their relative simplicity in both the structural feature and interaction patterns, still pose problems in theoretical modeling. The barrel-shaped supramolecular container cucurbit[8]uril (CB8) shows promising functionalities in various areas, e.g., catalysis and molecular recognition. It can stably coordinate a series of structurally diverse guests with high affinities. In this work, we examine the binding of seven commonly abused drugs to the CB8 host, aiming at providing a general picture of CB8−guest binding. Extensive sampling of the configurational space of these host−guest systems is performed, and the binding pathway and interaction patterns of CB8−guest complexes are investigated. A thorough comparison of widely used fixed-charge models for drug-like molecules is presented. Iterative refitting of the atomic charges suggests significant conformation dependence of charge generation. The initial model generated at the original conformation could be inaccurate for new conformations explored during conformational search, and the newly fitted charge set improves the prediction−experiment correlation significantly. Our investigations of the configurational space of CB8−drug complexes suggest that the host−guest interactions are more complex than expected. Despite the structural simplicities of these molecules, the conformational fluctuations of the host and the guest molecules and orientations of functional groups lead to the existence of an ensemble of binding modes. The insights of the binding thermodynamics, performance of fixed-charge models, and binding patterns of the CB8−guest systems are useful for studying and elucidating the binding mechanism of other host−guest complexes.

show abstract

Section: Introductionmentioning

confidence: 99%

A General Picture of Cucurbit[8]uril Host–Guest Binding

Sun

Huai²,

He³

et al. 2021

J. Chem. Inf. Model.

View full text Add to dashboard Cite

show abstract

“…The energy of the products in the reaction medium should be lower than the energy of the starting materials. The most basic thermodynamic processes are related to molecular properties such as tautomerisation, [10] sites of protonation and deprotonation, and calculation of partition coefficients [11,12] . Despite the simplicity of the concept, remarkably high levels of theory are needed as the energy difference between tautomers is often small, and even then the process might fail, due to the inadequate treatment of the solvent leading to levels of error (>1 kcal/mol) greater than the experimental difference [13] .…”

Section: Ex Vivo Chemical Reactivitymentioning

confidence: 99%

“…Reactions under thermodynamic control [10] Defining chemistry business rules for tautomer standardization that reproduce gas-phase energetic preferences Quantum mechanical calculations [11] Predicting partition coefficients Quantum mechanical calculations [11] Predicting dissociation constants Quantum mechanical calculations [12] Predicting partition coefficients and dissociation constants (SAMPL7 challenge)…”

Section: Rigid Reaction Templates and Neural Networkmentioning

confidence: 99%

Chemical Reactivity Prediction: Current Methods and Different Application Areas

Ertl

Gerebtzoff

Lewis

et al. 2022

Molecular Informatics

View full text Add to dashboard Cite

The ability to predict chemical reactivity of a molecule is highly desirable in drug discovery, both ex vivo (synthetic route planning, formulation, stability) and in vivo: metabolic reactions determine pharmacodynamics, pharmacokinetics and potential toxic effects, and early assessment of liabilities is vital to reduce attrition rates in later stages of development. Quantum mechanics offer a precise description of the interactions between electrons and orbitals in the breaking and forming of new bonds. Modern algorithms and faster computers have allowed the study of more complex systems in a punctual and accurate fashion, and answers for chemical questions around stability and reactivity can now be provided. Through machine learning, predictive models can be built out of descriptors derived from quantum mechanics and cheminformatics, even in the absence of experimental data to train on. In this article, current progress on computational reactivity prediction is reviewed: applications to problems in drug design, such as modelling of metabolism and covalent inhibition, are highlighted and unmet challenges are posed.

show abstract

“…1 Conceptually, logP measures the distribution of a molecule in neutral form between the octanol/water phases, while logD represents the distribution of both the ionized and un-ionized species of a molecule under a specific pH. 2 Thus, logD is a pH-dependent variant of logP, among which logD 7.4 is most relevant since the pH of plasma is 7.4. 3 Similarly, aqueous solubility (logS w ) is a variant of intrinsic solubility (logS 0 ) that considers ionization effects.…”

Section: ■ Introductionmentioning

confidence: 99%

“…Partition coefficient (log P ) and distribution coefficient (log D ) are two standard parameters for estimating lipophilicity . Conceptually, log P measures the distribution of a molecule in neutral form between the octanol/water phases, while log D represents the distribution of both the ionized and un-ionized species of a molecule under a specific pH . Thus, log D is a pH-dependent variant of log P , among which log D 7.4 is most relevant since the pH of plasma is 7.4 .…”

Section: Introductionmentioning

confidence: 99%

ALipSol: An Attention-Driven Mixture-of-Experts Model for Lipophilicity and Solubility Prediction

Wang

et al. 2022

J. Chem. Inf. Model.

View full text Add to dashboard Cite

Lipophilicity (logD) and aqueous solubility (logS w) play a central role in drug development. The accurate prediction of these properties remains to be solved due to data scarcity. Current methodologies neglect the intrinsic relationships between physicochemical properties and usually ignore the ionization effects. Here, we propose an attention-driven mixture-of-experts (MoE) model named ALipSol, which explicitly reproduces the hierarchy of task relationships. We adopt the principle of divide-and-conquer by breaking down the complex end point (logD or logS w) into simpler ones (acidic pK a, basic pK a, and logP) and allocating a specific expert network for each subproblem. Subsequently, we implement transfer learning to extract knowledge from related tasks, thus alleviating the dilemma of limited data. Additionally, we substitute the gating network with an attention mechanism to better capture the dynamic task relationships on a per-example basis. We adopt local fine-tuning and consensus prediction to further boost model performance. Extensive evaluation experiments verify the success of the ALipSol model, which achieves RMSE improvement of 8.04%, 2.49%, 8.57%, 12.8%, and 8.60% on the Lipop, ESOL, AqSolDB, external logD, and external logS data sets, respectively, compared with Attentive FP and the state-of-the-art in silico tools. In particular, our model yields more significant advantages (Welch’s t-test) for small training data, implying its high robustness and generalizability. The interpretability analysis proves that the atom contributions learned by ALipSol are more reasonable compared with the vanilla Attentive FP, and the substitution effects in benzene derivatives agreed well with empirical constants, revealing the potential of our model to extract useful patterns from data and provide guidance for lead optimization.

show abstract

SAMPL7 blind challenge: quantum–mechanical prediction of partition coefficients and acid dissociation constants for small drug-like molecules

Cited by 11 publications

References 65 publications

A General Picture of Cucurbit[8]uril Host–Guest Binding

A General Picture of Cucurbit[8]uril Host–Guest Binding

Chemical Reactivity Prediction: Current Methods and Different Application Areas

ALipSol: An Attention-Driven Mixture-of-Experts Model for Lipophilicity and Solubility Prediction

Contact Info

Product

Resources

About