Assessing the accuracy of octanol-water partition coefficient predictions in the SAMPL6 Part II log P Challenge

Llamol: a dynamic multi-conditional generative transformer for de novo molecular design

Generative models have demonstrated substantial promise in Natural Language Processing (NLP) and have found application in designing molecules, as seen in General Pretrained Transformer (GPT) models. In our efforts to develop such a tool for exploring the organic chemical space in search of potentially electro-active compounds, we present Llamol, a single novel generative transformer model based on the Llama 2 architecture, which was trained on a 12.5M superset of organic compounds drawn from diverse public sources. To allow for a maximum flexibility in usage and robustness in view of potentially incomplete data, we introduce Stochastic Context Learning (SCL) as a new training procedure. We demonstrate that the resulting model adeptly handles single- and multi-conditional organic molecule generation with up to four conditions, yet more are possible. The model generates valid molecular structures in SMILES notation while flexibly incorporating three numerical and/or one token sequence into the generative process, just as requested. The generated compounds are very satisfactory in all scenarios tested. In detail, we showcase the model's capability to utilize token sequences for conditioning, either individually or in combination with numerical properties, making Llamol a potent tool for de novo molecule design, easily expandable with new properties. SCIENTIFIC CONTRIBUTION: We developed a novel generative transformer model, Llamol, based on the Llama 2 architecture that was trained on a diverse set of 12.5 M organic compounds. It introduces Stochastic Context Learning (SCL) as a new training procedure, allowing for flexible and robust generation of valid organic molecules with up to multiple conditions that can be combined in various ways, making it a potent tool for de novo molecular design.

Blind prediction of toluene/water partition coefficients using COSMO-RS: results from the SAMPL9 challenge

Accurately predicting partition coefficients log P is crucial for reducing costs and accelerating drug design as it provides valuable information about the bioavailability, pharmacokinetics, and toxicity of different drug candidates. However, the performance of the existing methods is ambiguous, making it unclear whether these methods can be effectively utilized in drug discovery. To assess the performance of these methods, a series of SAMPL challenges have been conducted over the past few years, aiming to enable the development and validation of predictive models. In this study, we present two independent contributions to the SAMPL9 challenge for predicting the toluene/water partition coefficients for 16 molecules. Both submissions, A and B, use the COSMO-RS approach, albeit in slightly different procedures, to compute the transfer free energies from water to toluene of the molecules presented in the challenge, and consequently, their corresponding log P values. Based on the results, COSMO-RS submission A achieves the top position with an R2 value of 0.93 while it ranks second in terms of root-mean-square error (RMSE) with a value of 1.23 log P units. COSMO-RS submission B achieves an R2 value of 0.83 and an RMSE value of 1.48 log P units. Following the challenge, we predict the log P values using a neural network model, which was pre-trained on COSMO-RS data achieving an R2 of 0.92 and RMSE of 1.04 log P units. Compared to previous SAMPL challenges, all contributions displayed large deviations in predicting the toluene/water partition coefficient. These large deviations emphasize that further research is needed to develop accurate and reliable methods for modeling solvent effects on small molecule transfer-free energies.

Advances in Computational Approaches for Estimating Passive Permeability in Drug Discovery

Passive permeation of cellular membranes is a key feature of many therapeutics. The relevance of passive permeability spans all biological systems as they all employ biomembranes for compartmentalization. A variety of computational techniques are currently utilized and under active development to facilitate the characterization of passive permeability. These methods include lipophilicity relations, molecular dynamics simulations, and machine learning, which vary in accuracy, complexity, and computational cost. This review briefly introduces the underlying theories, such as the prominent inhomogeneous solubility diffusion model, and covers a number of recent applications. Various machine-learning applications, which have demonstrated good potential for high-volume, data-driven permeability predictions, are also discussed. Due to the confluence of novel computational methods and next-generation exascale computers, we anticipate an exciting future for computationally driven permeability predictions.

Recent Studies of Artificial Intelligence on In Silico Drug Absorption

Absorption is an important area of research in pharmacochemistry and drug development, because the drug has to be absorbed before any drug effects can occur. Furthermore, the ADMET (Absorption, Distribution, Metabolism, Excretion, and Toxicity) profile of drugs can be directly and considerably altered by modulating factors affecting absorption. Many drugs in development fail because of poor absorption. The research and continuous efforts of researchers in recent years have brought many successes and promises in drug absorption property prediction, especially in silico, which helps to reduce the time and cost significantly for screening undesirable drug candidates. In this report, we explicitly provide an overview of recent in silico studies on predicting absorption properties, especially from 2019 to the present, using artificial intelligence. Additionally, we have collected and investigated public databases that support absorption prediction research. On those grounds, we also proposed the challenges and development directions of absorption prediction in the future. We hope this review can provide researchers with valuable guidelines on absorption prediction to facilitate the development of newer approaches in drug discovery.

Adaptation of Empirical Methods to Predict the LogD of Triazine Macrocycles

Octanol/water partition coefficients guide drug design, but algorithms do not always accurately predict these values. For cationic triazine macrocycles that adopt a conserved folded shape in solution, common algorithms fall short. Here, the logD values for 12 macrocycles differing in amino acid choice were predicted and then measured experimentally. On average, AlogP, XlogP, and ChemAxon predictions deviate by 0.9, 2.8, and 3.9 log units, with XlogP overestimating lipophilicity and AlogP and ChemAxon underestimating lipophilicity. Importantly, however, a linear relationship (R2 > 0.98) exists between the values predicted by AlogP and the experimentally determined logD values, thus enabling more accurate predictions.

Prediction of toluene/water partition coefficients in the SAMPL9 blind challenge: assessment of machine learning and IEF-PCM/MST continuum solvation models

In recent years the use of partition systems other than the widely used biphasic n-octanol/water has received increased attention to gain insight into the molecular features that dictate the lipophilicity of compounds. Thus, the difference between n-octanol/water and toluene/water partition coefficients has proven to be a valuable descriptor to study the propensity of molecules to form intramolecular hydrogen bonds and exhibit chameleon-like properties that modulate solubility and permeability. In this context, this study reports the experimental toluene/water partition coefficients (log P_tol/w) for a series of 16 drugs that were selected as an external test set in the framework of the Statistical Assessment of the Modeling of Proteins and Ligands (SAMPL) blind challenge. This external set has been used by the computational community to calibrate their methods in the current edition (SAMPL9) of this contest. Furthermore, the study also investigates the performance of two computational strategies for the prediction of log P_tol/w. The first relies on the development of two machine learning (ML) models, which are built up by combining the selection of 11 molecular descriptors in conjunction with either the multiple linear regression (MLR) or the random forest regression (RFR) model to target a dataset of 252 experimental log P_tol/w values. The second consists of the parametrization of the IEF-PCM/MST continuum solvation model from B3LYP/6-31G(d) calculations to predict the solvation free energies of 163 compounds in toluene and benzene. The performance of the ML and IEF-PCM/MST models has been calibrated against external test sets, including the compounds that define the SAMPL9 log P_tol/w challenge. The results are used to discuss the merits and weaknesses of the two computational approaches.

Development and test of highly accurate endpoint free energy methods. 2: Prediction of logarithm of n-octanol-water partition coefficient (logP) for druglike molecules using MM-PBSA method

The logarithm of n-octanol-water partition coefficient (logP) is frequently used as an indicator of lipophilicity in drug discovery, which has substantial impacts on the absorption, distribution, metabolism, excretion, and toxicity of a drug candidate. Considering that the experimental measurement of the property is costly and time-consuming, it is of great importance to develop reliable prediction models for logP. In this study, we developed a transfer free energy-based logP prediction model-FElogP. FElogP is based on the simple principle that logP is determined by the free energy change of transferring a molecule from water to n-octanol. The underlying physical method to calculate transfer free energy is the molecular mechanics-Poisson Boltzmann surface area (MM-PBSA), thus this method is named as free energy-based logP (FElogP). The superiority of FElogP model was validated by a large set of 707 structurally diverse molecules in the ZINC database for which the measurement was of high quality. Encouragingly, FElogP outperformed several commonly-used QSPR or machine learning-based logP models, as well as some continuum solvation model-based methods. The root-mean-square error (RMSE) and Pearson correlation coefficient (R) between the predicted and measured values are 0.91 log units and 0.71, respectively, while the runner-up, the logP model implemented in OpenBabel had an RMSE of 1.13 log units and R of 0.67. Given the fact that FElogP was not parameterized against experimental logP directly, its excellent performance is likely to be expanded to arbitrary organic molecules covered by the general AMBER force fields.

The Chemical Space of Marine Antibacterials: Diphenyl Ethers, Benzophenones, Xanthones, and Anthraquinones

The emergence of multiresistant bacteria and the shortage of antibacterials in the drug pipeline creates the need to search for novel agents. Evolution drives the optimization of the structure of marine natural products to act as antibacterial agents. Polyketides are a vast and structurally diverse family of compounds that have been isolated from different marine microorganisms. Within the different polyketides, benzophenones, diphenyl ethers, anthraquinones, and xanthones have shown promising antibacterial activity. In this work, a dataset of 246 marine polyketides has been identified. In order to characterize the chemical space occupied by these marine polyketides, molecular descriptors and fingerprints were calculated. Molecular descriptors were analyzed according to the scaffold, and principal component analysis was performed to identify the relationships among the different descriptors. Generally, the identified marine polyketides are unsaturated, water-insoluble compounds. Among the different polyketides, diphenyl ethers tend to be more lipophilic and non-polar than the remaining classes. Molecular fingerprints were used to group the polyketides according to their molecular similarity into clusters. A total of 76 clusters were obtained, with a loose threshold for the Butina clustering algorithm, highlighting the large structural diversity of the marine polyketides. The large structural diversity was also evidenced by the visualization trees map assembled using the tree map (TMAP) unsupervised machine-learning method. The available antibacterial activity data were examined in terms of bacterial strains, and the activity data were used to rank the compounds according to their antibacterial potential. This potential ranking was used to identify the most promising compounds (four compounds) which can inspire the development of new structural analogs with better potency and absorption, distribution, metabolism, excretion, and toxicity (ADMET) properties.

The reconciliation between the experimental and calculated octanol-water partition coefficient of 1,2-dipalmitoyl-sn-glycero-3-phosphatidylcholine using atomistic molecular dynamics: an open question

The octanol-water partition coefficient of 1,2-dipalmitoyl-sn-glycero-3-phosphatidylcholine (DPPC) was investigated using atomistic molecular dynamics simulations via thermodynamic integration and multistate Bennett acceptance ratio methods. The GAFF and CHARMM36 force fields were used with six water models widely used in molecular dynamics simulations. The OPC4 water model provided the best agreement with the experimental octanol-water partition coefficient of DPPC using the two force fields. However, there is still plenty of room for improvement in water models with correct estimation of surface tension that uses better and suitable non-bonded interaction parameters between water-water and water-DPPC. The Gibbs free energy of transferring DPPC from octanol to water phase was calculated to be 19.8 ± 0.3 and 20.2 ± 0.3 kcal mol-1, giving a partition coefficient of 14.5 ± 0.4 and 14.8 ± 0.3 for the GAFF and CHARMM36 force fields, respectively. This study reinforces the importance of developing new water models that reproduce experimental surface tensions to reconcile the water-water and water-DPPC non-bonded interactions and the existing discrepancy between experimental measurements of amphiphilic molecules that are important in many areas of scientific applications and industry such as biophysics, surfactant, colloids, membranes, medicine, nanotechnology, and food and pharmaceutical industries, and so on. It raises two important open questions: Is the experimental octanol-water partition coefficient of DPPC reliable? Or is its calculation accurate using the OPC4 water model? With respect to the experimental measurements, there may be non-treated aspects such as the formation of aggregates in aqueous phase and limit of detection of the applied method. And, in the calculation, some effects are not possible to be considered in a correct way or viable time such as calculating quantum effects, sampling all conformations, considering phase transitions, and correctly evaluating the intermolecular forces to estimate an accurate surface tension.Communicated by Ramaswamy H. Sarma.

Machine Learning for Fast, Quantum Mechanics-Based Approximation of Drug Lipophilicity

Lipophilicity, as measured by the partition coefficient between octanol and water (log P), is a key parameter in early drug discovery research. However, measuring log P experimentally is difficult for specific compounds and log P ranges. The resulting lack of reliable experimental data impedes development of accurate in silico models for such compounds. In certain discovery projects at Novartis focused on such compounds, a quantum mechanics (QM)-based tool for log P estimation has emerged as a valuable supplement to experimental measurements and as a preferred alternative to existing empirical models. However, this QM-based approach incurs a substantial computational cost, limiting its applicability to small series and prohibiting quick, interactive ideation. This work explores a set of machine learning models (Random Forest, Lasso, XGBoost, Chemprop, and Chemprop3D) to learn calculated log P values on both a public data set and an in-house data set to obtain a computationally affordable, QM-based estimation of drug lipophilicity. The message-passing neural network model Chemprop emerged as the best performing model with mean absolute errors of 0.44 and 0.34 log units for scaffold split test sets of the public and in-house data sets, respectively. Analysis of learning curves suggests that a further decrease in the test set error can be achieved by increasing the training set size. While models directly trained on experimental data perform better at approximating experimentally determined log P values than models trained on calculated values, we discuss the potential advantages of using calculated log P values going beyond the limits of experimental quantitation. We analyze the impact of the data set splitting strategy and gain insights into model failure modes. Potential use cases for the presented models include pre-screening of large compound collections and prioritization of compounds for full QM calculations.

Fast Prediction of Lipophilicity of Organofluorine Molecules: Deep Learning-Derived Polarity Characters and Experimental Tests

Fast and accurate estimation of lipophilicity for organofluorine molecules is in great demand for accelerating drug and materials discovery. A lipophilicity data set of organofluorine molecules (OFL data set), containing 1907 samples, is constructed through density functional theory (DFT) calculations and experimental measurements. An efficient and interpretable model, called PoLogP, is developed to predict the n-octanol/water partition coefficient, log P_o/w, of organofluorine molecules on the basis of the descriptors of polarization, which is a combination of polarity descriptors, including the molecular polarity index and molecular polarizability (α), and hydrogen bond (HBs) index, consisting of the number of donors (N_HB^D) and acceptors (N_HB^A and N_HB-F^A). The present PoLogP with a combination of polarity descriptors is demonstrated to perform better than the dipole moment (μ) alone for the F-contained molecules. With the aid of a multilevel attention graph convolutional neural network model, the fast generation of polarity descriptors of organofluorine molecules could be achieved with the DFT accuracy based only on a topological molecular graph structure. The performance of PoLogP is further validated on synthesized organofluorine molecules and 2626 non-fluorinated molecules with satisfactory accuracy, highlighting the potential usage of PoLogP in high-throughput screening of the functional molecules with the desired solubility in various solvent media.

Predicting octanol/water partition coefficients and pKa for the SAMPL7 challenge using the SM12, SM8 and SMD solvation models

Blind predictions of octanol/water partition coefficients and pKa at 298.15 K for 22 drug-like compounds were made for the SAMPL7 challenge. Octanol/water partition coefficients were predicted from solvation free energies computed using electronic structure calculations with the SM12, SM8 and SMD solvation models. Within these calculations we compared the use of gas- and solution-phase optimized geometries of the solute. Based on these calculations we found that in general the use of solution phase-optimized geometries increases the affinity of the solutes for water as compared to octanol, with the use of gas-phase optimized geometries resulting in the better agreement with experiment. The pKa is computed using the direct approach, scaled solvent-accessible surface model, and the inclusion of an explicit water molecule, where the latter two methods have previously been shown to offer improved predictions as compared to the direct approach. We find that the use of an explicit water molecule provides superior predictions, and that the predicted macroscopic pKa is sensitive to the employed microstates.

Implementation and Optimization of the Embedded Cluster Reference Interaction Site Model with Atomic Charges

In this work, we implemented the embedded cluster reference interaction site model (EC-RISM) originally developed by Kloss, Heil, and Kast (J. Phys. Chem. B2008, 112, 4337–4343). This method combines quantum mechanical calculations with the 3D reference interaction site model (3D-RISM). Numerous options, such as buffer, grid space, basis set, charge model, water model, closure relation, and so forth, were investigated to find the best settings. Additionally, the small point charges, which are derived from the solvent distribution from the 3D-RISM solution to represent the solvent in the QM calculation, were neglected to reduce the overhead without the loss of accuracy. On the MNSOL[a], MNSOL, and FreeSolv databases, our implemented and optimized method provides solvation free energies in water with 5.70, 6.32, and 6.44 kJ/mol root-mean-square deviations, respectively, but with different settings, 5.22, 6.08, and 6.63 kJ/mol can also be achieved. Only solvent models containing fitting parameters, like COSMO-RS and EC-RISM with universal correction and directly used electrostatic potential, perform better than our EC-RISM implementation with atomic charges.

Comparison of logP and logD correction models trained with public and proprietary data sets

In drug discovery, partition and distribution coefficients, logP and logD for octanol/water, are widely used as metrics of the lipophilicity of molecules, which in turn have a strong influence on the bioactivity and bioavailability of potential drugs. There are a variety of established methods, mostly fragment or atom-based, to calculate logP while logD prediction generally relies on calculated logP and pKa for the estimation of neutral and ionized populations at a given pH. Algorithms such as ClogP have limitations generally leading to systematic errors for chemically related molecules while pKa estimation is generally more difficult due to the interplay of electronic, inductive and conjugation effects for ionizable moieties. We propose an integrated machine learning QSAR modeling approach to predict logD by training the model with experimental data while using ClogP and pKa predicted by commercial software as model descriptors. By optimizing the loss function for the ClogD calculated by the software, we build a correction model that incorporates both descriptors from the software and available experimental logD data. Additionally, we calculate logP from the logD model using the software predicted pKa's. Here, we have trained models using publicly or commercial available logD data to show that this approach can improve on commercial software predictions of lipophilicity. When applied to other logD data sets, this approach extends the domain of applicability of logD and logP predictions over commercial software. Performance of these models favorably compare with models built with a larger set of proprietary logD data.

SAMPL7 protein-ligand challenge: A community-wide evaluation of computational methods against fragment screening and pose-prediction

A novel crystallographic fragment screening data set was generated and used in the SAMPL7 challenge for protein-ligands. The SAMPL challenges prospectively assess the predictive power of methods involved in computer-aided drug design. Application of various methods to fragment molecules are now widely used in the search for new drugs. However, there is little in the way of systematic validation specifically for fragment-based approaches. We have performed a large crystallographic high-throughput fragment screen against the therapeutically relevant second bromodomain of the Pleckstrin-homology domain interacting protein (PHIP2) that revealed 52 different fragments bound across 4 distinct sites, 47 of which were bound to the pharmacologically relevant acetylated lysine (Kac) binding site. These data were used to assess computational screening, binding pose prediction and follow-up enumeration. All submissions performed randomly for screening. Pose prediction success rates (defined as less than 2 Å root mean squared deviation against heavy atom crystal positions) ranged between 0 and 25% and only a very few follow-up compounds were deemed viable candidates from a medicinal-chemistry perspective based on a common molecular descriptors analysis. The tight deadlines imposed during the challenge led to a small number of submissions suggesting that the accuracy of rapidly responsive workflows remains limited. In addition, the application of these methods to reproduce crystallographic fragment data still appears to be very challenging. The results show that there is room for improvement in the development of computational tools particularly when applied to fragment-based drug design.

AB-DB: Force-Field parameters, MD trajectories, QM-based data, and Descriptors of Antimicrobials

Antibiotic resistance is a major threat to public health. The development of chemo-informatic tools to guide medicinal chemistry campaigns in the efficint design of antibacterial libraries is urgently needed. We present AB-DB, an open database of all-atom force-field parameters, molecular dynamics trajectories, quantum-mechanical properties, and curated physico-chemical descriptors of antimicrobial compounds. We considered more than 300 molecules belonging to 25 families that include the most relevant antibiotic classes in clinical use, such as β-lactams and (fluoro)quinolones, as well as inhibitors of key bacterial proteins. We provide traditional descriptors together with properties obtained with Density Functional Theory calculations. Noteworthy, AB-DB contains less conventional descriptors extracted from μs-long molecular dynamics simulations in explicit solvent. In addition, for each compound we make available force-field parameters for the major micro-species at physiological pH. With the rise of multi-drug-resistant pathogens and the consequent need for novel antibiotics, inhibitors, and drug re-purposing strategies, curated databases containing reliable and not straightforward properties facilitate the integration of data mining and statistics into the discovery of new antimicrobials.

Measurement(s)	molecular physical property analysis objective
Technology Type(s)	Computer Modeling

Quantum Chemical Calculation and Evaluation of Partition Coefficients for Classical and Emerging Environmentally Relevant Organic Compounds

Octanol/water (K_OW), octanol/air (K_OA), and hexadecane/air (K_HdA) partition coefficients are calculated for 67 organic compounds of environmental concern using computational chemistry. The extended CRENSO workflow applied here includes the calculation of extensive conformer ensembles with semiempirical methods and refinement through density functional theory, taking into account solvation models, especially COSMO-RS, and thermostatistical contributions. This approach is particularly advantageous for describing large and nonrigid molecules. With regard to K_OW and K_HdA, one can refer to many experimental data from direct and indirect measurement methods, and very good matches with results from our quantum chemical workflow are evident. In the case of the K_OA values, however, good matches are only obtained for the experimentally determined values. Larger systematic deviations between data computed here and available, nonexperimental quantitative structure-activity relationship literature data occur in particular for phthalic acid esters and organophosphate esters. From a critical analysis of the coefficients calculated in this work and comparison with available literature data, we conclude that the presented quantum chemical composite approach is the most powerful so far for calculating reliable partition coefficients because all physical contributions to the conformational free energy are considered and the structure ensembles for the two phases are generated independently and consistently.

Molecular characteristics of several drugs evaluated from solvent/water partition measurements: Solvation parameters and intramolecular hydrogen bond indicator

A wide set of well-known drugs, most of them included in the Abraham´s reference database, covering a wide variety of chemical structures and therapeutical functionalities were chosen in order to determine some molecular properties from solvent/water partition measurements. Partition data from aqueous solutions and four different solvents (n-dodecane, toluene, chloroform and n-octanol) were measured and reported. From them, Abraham´s molecular descriptors of selected compounds (A, B and S, accounting for hydrogen bond donor, hydrogen bond acceptor and dipolarity/polaritzability, respectively) were estimated. A and B values derived from the experimental measurements strongly agree with the tabulated ones showing the suitability of the used procedure to achieve reliable values for new molecules. However, obtained S values differ from those previously reported for several compounds. Moreover, values for a new indicator of the propensity to form intramolecular hydrogen bonds (Δlog P_oct-tol) were estimated from the experimental data and also calculated according to both, the Abraham´s model and the molecular structures (SMD). The quality of both series of calculated descriptors was evaluated by contrast with the experimental values and satisfactory results were obtained in both instances. Thus, the Abraham´s way is useful when molecular descriptors are available but very good estimations can be achieved by SMD, which only requires the drug´s molecular structure.

Automated high throughput pKa and distribution coefficient measurements of pharmaceutical compounds for the SAMPL8 blind prediction challenge

The goal of the Statistical Assessment of the Modeling of Proteins and Ligands (SAMPL) challenge is to improve the accuracy of current computational models to estimate free energy of binding, deprotonation, distribution and other associated physical properties that are useful for the design of new pharmaceutical products. New experimental datasets of physicochemical properties provide opportunities for prospective evaluation of computational prediction methods. Here, aqueous pK_a and a range of bi-phasic logD values for a variety of pharmaceutical compounds were determined through a streamlined automated process to be utilized in the SAMPL8 physical property challenge. The goal of this paper is to provide an in-depth review of the experimental methods utilized to create a comprehensive data set for the blind prediction challenge. The significance of this work involves the use of high throughput experimentation equipment and instrumentation to produce acid dissociation constants for twenty-three drug molecules, as well as distribution coefficients for eleven of those molecules.

Predicting octanol/water partition coefficients using molecular simulation for the SAMPL7 challenge: comparing the use of neat and water saturated 1-octanol

Blind predictions of octanol/water partition coefficients at 298.15 K for 22 drug-like compounds were made for the SAMPL7 challenge. The octanol/water partition coefficients were predicted using solvation free energies computed using molecular dynamics simulations, wherein we considered the use of both pure and water-saturated 1-octanol to model the octanol-rich phase. Water and 1-octanol were modeled using TIP4P and TrAPPE-UA, respectively, which have been shown to well reproduce the experimental mutual solubility, and the solutes were modeled using GAFF. After the close of the SAMPL7 challenge, we additionally made predictions using TIP4P/2005 water. We found that the predictions were sensitive to the choice of water force field. However, the effect of water in the octanol-rich phase was found to be even more significant and non-negligible. The effect of inclusion of water was additionally sensitive to the chemical structure of the solute.

Multiple linear regression models for predicting the n‑octanol/water partition coefficients in the SAMPL7 blind challenge

A multiple linear regression model called MLR-3 is used for predicting the experimental n-octanol/water partition coefficient (log PN) of 22 N-sulfonamides proposed by the organizers of the SAMPL7 blind challenge. The MLR-3 method was trained with 82 molecules including drug-like sulfonamides and small organic molecules, which resembled the main functional groups present in the challenge dataset. Our model, submitted as "TFE-MLR", presented a root-mean-square error of 0.58 and mean absolute error of 0.41 in log P units, accomplishing the highest accuracy, among empirical methods and also in all submissions based on the ranked ones. Overall, the results support the appropriateness of multiple linear regression approach MLR-3 for computing the n-octanol/water partition coefficient in sulfonamide-bearing compounds. In this context, the outstanding performance of empirical methodologies, where 75% of the ranked submissions achieved root-mean-square errors < 1 log P units, support the suitability of these strategies for obtaining accurate and fast predictions of physicochemical properties as partition coefficients of bioorganic compounds.

Prediction of n-octanol/water partition coefficients and acidity constants (pKa) in the SAMPL7 blind challenge with the IEFPCM-MST model

Within the scope of SAMPL7 challenge for predicting physical properties, the Integral Equation Formalism of the Miertus-Scrocco-Tomasi (IEFPCM/MST) continuum solvation model has been used for the blind prediction of n-octanol/water partition coefficients and acidity constants of a set of 22 and 20 sulfonamide-containing compounds, respectively. The log P and pK_a were computed using the B3LPYP/6-31G(d) parametrized version of the IEFPCM/MST model. The performance of our method for partition coefficients yielded a root-mean square error of 1.03 (log P units), placing this method among the most accurate theoretical approaches in the comparison with both globally (rank 8th) and physical (rank 2nd) methods. On the other hand, the deviation between predicted and experimental pK_a values was 1.32 log units, obtaining the second best-ranked submission. Though this highlights the reliability of the IEFPCM/MST model for predicting the partitioning and the acid dissociation constant of drug-like compounds compound, the results are discussed to identify potential weaknesses and improve the performance of the method.

Energy-entropy prediction of octanol-water logP of SAMPL7 N-acyl sulfonamide bioisosters

Partition coefficients quantify a molecule's distribution between two immiscible liquid phases. While there are many methods to compute them, there is not yet a method based on the free energy of each system in terms of energy and entropy, where entropy depends on the probability distribution of all quantum states of the system. Here we test a method in this class called Energy Entropy Multiscale Cell Correlation (EE-MCC) for the calculation of octanol-water logP values for 22 N-acyl sulfonamides in the SAMPL7 Physical Properties Challenge (Statistical Assessment of the Modelling of Proteins and Ligands). EE-MCC logP values have a mean error of 1.8 logP units versus experiment and a standard error of the mean of 1.0 logP units for three separate calculations. These errors are primarily due to getting sufficiently converged energies to give accurate differences of large numbers, particularly for the large-molecule solvent octanol. However, this is also an issue for entropy, and approximations in the force field and MCC theory also contribute to the error. Unique to MCC is that it explains the entropy contributions over all the degrees of freedom of all molecules in the system. A gain in orientational entropy of water is the main favourable entropic contribution, supported by small gains in solute vibrational and orientational entropy but offset by unfavourable changes in the orientational entropy of octanol, the vibrational entropy of both solvents, and the positional and conformational entropy of the solute.

Predicting partition coefficients for the SAMPL7 physical property challenge using the ClassicalGSG method

The prediction of [Formula: see text] values is one part of the statistical assessment of the modeling of proteins and ligands (SAMPL) blind challenges. Here, we use a molecular graph representation method called Geometric Scattering for Graphs (GSG) to transform atomic attributes to molecular features. The atomic attributes used here are parameters from classical molecular force fields including partial charges and Lennard-Jones interaction parameters. The molecular features from GSG are used as inputs to neural networks that are trained using a "master" dataset comprised of over 41,000 unique [Formula: see text] values. The specific molecular targets in the SAMPL7 [Formula: see text] prediction challenge were unique in that they all contained a sulfonyl moeity. This motivated a set of ClassicalGSG submissions where predictors were trained on different subsets of the master dataset that are filtered according to chemical types and/or the presence of the sulfonyl moeity. We find that our ranked prediction obtained 5th place with an RMSE of 0.77 [Formula: see text] units and an MAE of 0.62, while one of our non-ranked predictions achieved first place among all submissions with an RMSE of 0.55 and an MAE of 0.44. After the conclusion of the challenge we also examined the performance of open-source force field parameters that allow for an end-to-end [Formula: see text] predictor model: General AMBER Force Field (GAFF), Universal Force Field (UFF), Merck Molecular Force Field 94 (MMFF94) and Ghemical. We find that ClassicalGSG models trained with atomic attributes from MMFF94 can yield more accurate predictions compared to those trained with CGenFF atomic attributes.

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

The physicochemical properties of a drug molecule determine the therapeutic effectiveness of the drug. Thus, the development of fast and accurate theoretical approaches for the prediction of such properties is inevitable. The participation to the SAMPL7 challenge is based on the estimation of logP coefficients and pK_a values of small drug-like sulfonamide derivatives. Thereby, quantum mechanical calculations were carried out in order to calculate the free energy of solvation and the transfer energy of 22 drug-like compounds in different environments (water and n-octanol) by employing the SMD solvation model. For logP calculations, we studied eleven different methodologies to calculate the transfer free energies, the lowest RMSE value was obtained for the M06L/def2-TZVP//M06L/def2-SVP level of theory. On the other hand, we employed an isodesmic reaction scheme within the macro pK_a framework; this was based on selecting reference molecules similar to the SAMPL7 challenge molecules. Consequently, highly well correlated pK_a values were obtained with the M062X/6-311+G(2df,2p)//M052X/6-31+G(d,p) level of theory.

Evaluation of log P, pKa, and log D predictions from the SAMPL7 blind challenge

The Statistical Assessment of Modeling of Proteins and Ligands (SAMPL) challenges focuses the computational modeling community on areas in need of improvement for rational drug design. The SAMPL7 physical property challenge dealt with prediction of octanol-water partition coefficients and pK_a for 22 compounds. The dataset was composed of a series of N-acylsulfonamides and related bioisosteres. 17 research groups participated in the log P challenge, submitting 33 blind submissions total. For the pK_a challenge, 7 different groups participated, submitting 9 blind submissions in total. Overall, the accuracy of octanol-water log P predictions in the SAMPL7 challenge was lower than octanol-water log P predictions in SAMPL6, likely due to a more diverse dataset. Compared to the SAMPL6 pK_a challenge, accuracy remains unchanged in SAMPL7. Interestingly, here, though macroscopic pK_a values were often predicted with reasonable accuracy, there was dramatically more disagreement among participants as to which microscopic transitions produced these values (with methods often disagreeing even as to the sign of the free energy change associated with certain transitions), indicating far more work needs to be done on pK_a prediction methods.

Overview of the SAMPL6 pKa challenge: evaluating small molecule microscopic and macroscopic pKa predictions

The prediction of acid dissociation constants (pKa) is a prerequisite for predicting many other properties of a small molecule, such as its protein-ligand binding affinity, distribution coefficient (log D), membrane permeability, and solubility. The prediction of each of these properties requires knowledge of the relevant protonation states and solution free energy penalties of each state. The SAMPL6 pKa Challenge was the first time that a separate challenge was conducted for evaluating pKa predictions as part of the Statistical Assessment of Modeling of Proteins and Ligands (SAMPL) exercises. This challenge was motivated by significant inaccuracies observed in prior physical property prediction challenges, such as the SAMPL5 log D Challenge, caused by protonation state and pKa prediction issues. The goal of the pKa challenge was to assess the performance of contemporary pKa prediction methods for drug-like molecules. The challenge set was composed of 24 small molecules that resembled fragments of kinase inhibitors, a number of which were multiprotic. Eleven research groups contributed blind predictions for a total of 37 pKa distinct prediction methods. In addition to blinded submissions, four widely used pKa prediction methods were included in the analysis as reference methods. Collecting both microscopic and macroscopic pKa predictions allowed in-depth evaluation of pKa prediction performance. This article highlights deficiencies of typical pKa prediction evaluation approaches when the distinction between microscopic and macroscopic pKas is ignored; in particular, we suggest more stringent evaluation criteria for microscopic and macroscopic pKa predictions guided by the available experimental data. Top-performing submissions for macroscopic pKa predictions achieved RMSE of 0.7-1.0 pKa units and included both quantum chemical and empirical approaches, where the total number of extra or missing macroscopic pKas predicted by these submissions were fewer than 8 for 24 molecules. A large number of submissions had RMSE spanning 1-3 pKa units. Molecules with sulfur-containing heterocycles or iodo and bromo groups were less accurately predicted on average considering all methods evaluated. For a subset of molecules, we utilized experimentally-determined microstates based on NMR to evaluate the dominant tautomer predictions for each macroscopic state. Prediction of dominant tautomers was a major source of error for microscopic pKa predictions, especially errors in charged tautomers. The degree of inaccuracy in pKa predictions observed in this challenge is detrimental to the protein-ligand binding affinity predictions due to errors in dominant protonation state predictions and the calculation of free energy corrections for multiple protonation states. Underestimation of ligand pKa by 1 unit can lead to errors in binding free energy errors up to 1.2 kcal/mol. The SAMPL6 pKa Challenge demonstrated the need for improving pKa prediction methods for drug-like molecules, especially for challenging moieties and multiprotic molecules.

SAMPL7 Host-Guest Challenge Overview: assessing the reliability of polarizable and non-polarizable methods for binding free energy calculations

The SAMPL challenges focus on testing and driving progress of computational methods to help guide pharmaceutical drug discovery. However, assessment of methods for predicting binding affinities is often hampered by computational challenges such as conformational sampling, protonation state uncertainties, variation in test sets selected, and even lack of high quality experimental data. SAMPL blind challenges have thus frequently included a component focusing on host-guest binding, which removes some of these challenges while still focusing on molecular recognition. Here, we report on the results of the SAMPL7 blind prediction challenge for host-guest affinity prediction. In this study, we focused on three different host-guest categories-a familiar deep cavity cavitand series which has been featured in several prior challenges (where we examine binding of a series of guests to two hosts), a new series of cyclodextrin derivatives which are monofunctionalized around the rim to add amino acid-like functionality (where we examine binding of two guests to a series of hosts), and binding of a series of guests to a new acyclic TrimerTrip host which is related to previous cucurbituril hosts. Many predictions used methods based on molecular simulations, and overall success was mixed, though several methods stood out. As in SAMPL6, we find that one strategy for achieving reasonable accuracy here was to make empirical corrections to binding predictions based on previous data for host categories which have been studied well before, though this can be of limited value when new systems are included. Additionally, we found that alchemical free energy methods using the AMOEBA polarizable force field had considerable success for the two host categories in which they participated. The new TrimerTrip system was also found to introduce some sampling problems, because multiple conformations may be relevant to binding and interconvert only slowly. Overall, results in this challenge tentatively suggest that further investigation of polarizable force fields for these challenges may be warranted.

Non-equilibrium approach for binding free energies in cyclodextrins in SAMPL7: force fields and software

In the current work we report on our participation in the SAMPL7 challenge calculating absolute free energies of the host–guest systems, where 2 guest molecules were probed against 9 hosts-cyclodextrin and its derivatives. Our submission was based on the non-equilibrium free energy calculation protocol utilizing an averaged consensus result from two force fields (GAFF and CGenFF). The submitted prediction achieved accuracy of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${1.38}\,\hbox {kcal}/\hbox {mol}$$\end{document}1.38kcal/mol in terms of the unsigned error averaged over the whole dataset. Subsequently, we further report on the underlying reasons for discrepancies between our calculations and another submission to the SAMPL7 challenge which employed a similar methodology, but disparate ligand and water force fields. As a result we have uncovered a number of issues in the dihedral parameter definition of the GAFF 2 force field. In addition, we identified particular cases in the molecular topologies where different software packages had a different interpretation of the same force field. This latter observation might be of particular relevance for systematic comparisons of molecular simulation software packages. The aforementioned factors have an influence on the final free energy estimates and need to be considered when performing alchemical calculations.

Matched Molecular Series Analysis for ADME Property Prediction

Generation and prioritization of new molecules are the most central part of the drug design process. Matched molecular series analysis (MMSA) has recently been proposed as a formal approach that captures both of these key elements of design. In order to better understand the power of MMSA and its specific limitations, we here evaluate its performance as an ADME property prediction tool. We use four large and diverse inhouse data sets, logD, microsomal clearance, CYP2C9, and CYP3A4 inhibition. MMSA follows the concept of parallel structure-activity relationship (SAR), where if two identical substituent series on different scaffolds show similarity in their property profiles, SAR from one series can be transferred to the other series. We test four different similarity metrics to identify pairs of molecular series where information can be transferred. We find that the best prediction performance is achieved by a combination of centered root-mean-square deviation (cRMSD) and a network score approach previously published by Keefer et al. However, cRMSD alone strikes the best balance between accuracy and the number of predictions that can be made. We identify statistical metrics that allow estimating when MMSA predictions will work, similar to the well-known applicability domain concept in machine learning. MMSA achieves a prediction accuracy that is comparable to a standard machine-learning model and matched molecular pair analysis. In contrast to machine learning, however, it is very easy to understand where MMSA predictions are coming from. Finally, to prospectively test the power of MMSA, we retested compounds that were strong outliers in the initial predictions and show how the MMSA model can help to identify erroneous data points.

Prediction of octanol-water partition coefficients for the SAMPL6-[Formula: see text] molecules using molecular dynamics simulations with OPLS-AA, AMBER and CHARMM force fields

All-atom molecular dynamics simulations with stratified alchemical free energy calculations were used to predict the octanol-water partition coefficient [Formula: see text] of eleven small molecules as part of the SAMPL6-[Formula: see text] blind prediction challenge using four different force field parametrizations: standard OPLS-AA with transferable charges, OPLS-AA with non-transferable CM1A charges, AMBER/GAFF, and CHARMM/CGenFF. Octanol parameters for OPLS-AA, GAFF and CHARMM were validated by comparing the density as a function of temperature, the chemical potential, and the hydration free energy to experimental values. The partition coefficients were calculated from the solvation free energy for the compounds in water and pure ("dry") octanol or "wet" octanol with 27 mol% water dissolved. Absolute solvation free energies were computed by thermodynamic integration (TI) and the multistate Bennett acceptance ratio with uncorrelated samples from data generated by an established protocol using 5-ns windowed alchemical free energy perturbation (FEP) calculations with the Gromacs molecular dynamics package. Equilibration of sets of FEP simulations was quantified by a new measure of convergence based on the analysis of forward and time-reversed trajectories. The accuracy of the [Formula: see text] predictions was assessed by descriptive statistical measures such as the root mean square error (RMSE) of the data set compared to the experimental values. Discarding the first 1 ns of each 5-ns window as an equilibration phase had a large effect on the GAFF data, where it improved the RMSE by up to 0.8 log units, while the effect for other data sets was smaller or marginally worsened the agreement. Overall, CGenFF gave the best prediction with RMSE 1.2 log units, although for only eight molecules because the current CGenFF workflow for Gromacs does not generate files for certain halogen-containing compounds. Over all eleven compounds, GAFF gave an RMSE of 1.5. The effect of using a mixed water/octanol solvent slightly decreased the accuracy for CGenFF and GAFF and slightly increased it for OPLS-AA. The GAFF and OPLS-AA results displayed a systematic error where molecules were too hydrophobic whereas CGenFF appeared to be more balanced, at least on this small data set.

Octanol-water partition coefficient measurements for the SAMPL6 blind prediction challenge

Partition coefficients describe the equilibrium partitioning of a single, defined charge state of a solute between two liquid phases in contact, typically a neutral solute. Octanol-water partition coefficients ([Formula: see text]), or their logarithms (log P), are frequently used as a measure of lipophilicity in drug discovery. The partition coefficient is a physicochemical property that captures the thermodynamics of relative solvation between aqueous and nonpolar phases, and therefore provides an excellent test for physics-based computational models that predict properties of pharmaceutical relevance such as protein-ligand binding affinities or hydration/solvation free energies. The SAMPL6 Part II octanol-water partition coefficient prediction challenge used a subset of kinase inhibitor fragment-like compounds from the SAMPL6 [Formula: see text] prediction challenge in a blind experimental benchmark. Following experimental data collection, the partition coefficient dataset was kept blinded until all predictions were collected from participating computational chemistry groups. A total of 91 submissions were received from 27 participating research groups. This paper presents the octanol-water log P dataset for this SAMPL6 Part II partition coefficient challenge, which consisted of 11 compounds (six 4-aminoquinazolines, two benzimidazole, one pyrazolo[3,4-d]pyrimidine, one pyridine, one 2-oxoquinoline substructure containing compounds) with log P values in the range of 1.95-4.09. We describe the potentiometric log P measurement protocol used to collect this dataset using a Sirius T3, discuss the limitations of this experimental approach, and share suggestions for future log P data collection efforts for the evaluation of computational methods.

Prediction of the n-octanol/water partition coefficients in the SAMPL6 blind challenge from MST continuum solvation calculations

The IEFPCM/MST continuum solvation model is used for the blind prediction of n-octanol/water partition of a set of 11 fragment-like small molecules within the SAMPL6 Part II Partition Coefficient Challenge. The partition coefficient of the neutral species (log P) was determined using an extended parametrization of the B3LYP/6-31G(d) version of the Miertus-Scrocco-Tomasi continuum solvation model in n-octanol. Comparison with the experimental data provided for partition coefficients yielded a root-mean square error (rmse) of 0.78 (log P units), which agrees with the accuracy reported for our method (rmse = 0.80) for nitrogen-containing heterocyclic compounds. Out of the 91 sets of log P values submitted by the participants, our submission is within those with an rmse < 1 and among the four best ranked physical methods. The largest errors involve three compounds: two with the largest positive deviations (SM13 and SM08), and one with the largest negative deviations (SM15). Here we report the potentiometric determination of the log P for SM13, leading to a value of 3.62 ± 0.02, which is in better agreement with most empirical predictions than the experimental value reported in SAMPL6. In addition, further inclusion of several conformations for SM08 significantly improved our results. Inclusion of these refinements led to an overall error of 0.51 (log P units), which supports the reliability of the IEFPCM/MST model for predicting the partitioning of neutral compounds.