JPlogP: an improved logP predictor trained using predicted data

The openOCHEM consensus model is the best-performing open-source predictive model in the First EUOS/SLAS joint compound solubility challenge

The EUOS/SLAS challenge aimed to facilitate the development of reliable algorithms to predict the aqueous solubility of small molecules using experimental data from 100 K compounds. In total, hundred teams took part in the challenge to predict low, medium and highly soluble compounds as measured by the nephelometry assay. This article describes the winning model, which was developed using the publicly available Online CHEmical database and Modeling environment (OCHEM) available on the website https://ochem.eu/article/27. We describe in detail the assumptions and steps used to select methods, descriptors and strategy which contributed to the winning solution. In particular we show that consensus based on 28 models calculated using descriptor-based and representation learning methods allowed us to obtain the best score, which was higher than those based on individual approaches or consensus models developed using each individual approach. A combination of diverse models allowed us to decrease both bias and variance of individual models and to calculate the highest score. The model based on Transformer CNN contributed the best individual score thus highlighting the power of Natural Language Processing (NLP) methods. The inclusion of information about aleatoric uncertainty would be important to better understand and use the challenge data by the contestants.

Bridging the Gap between Differential Mobility, Log S, and Log P Using Machine Learning and SHAP Analysis

Aqueous solubility, log S, and the water-octanol partition coefficient, log P, are physicochemical properties that are used to screen the viability of drug candidates and to estimate mass transport in the environment. In this work, differential mobility spectrometry (DMS) experiments performed in microsolvating environments are used to train machine learning (ML) frameworks that predict the log S and log P of various molecule classes. In lieu of a consistent source of experimentally measured log S and log P values, the OPERA package was used to evaluate the aqueous solubility and hydrophobicity of 333 analytes. With ion mobility/DMS data (e.g., CCS, dispersion curves) as input, we used ML regressors and ensemble stacking to derive relationships with a high degree of explainability, as assessed via SHapley Additive exPlanations (SHAP) analysis. The DMS-based regression models returned scores of R² = 0.67 and RMSE = 1.03 ± 0.10 for log S predictions and R² = 0.67 and RMSE = 1.20 ± 0.10 for log P after 5-fold random cross-validation. SHAP analysis reveals that the regressors strongly weighted gas-phase clustering in log P correlations. The addition of structural descriptors (e.g., # of aromatic carbons) improved log S predictions to yield RMSE = 0.84 ± 0.07 and R² = 0.78. Similarly, log P predictions using the same data resulted in an RMSE of 0.83 ± 0.04 and R² = 0.84. The SHAP analysis of log P models highlights the need for additional experimental parameters describing hydrophobic interactions. These results were achieved with a smaller dataset (333 instances) and minimal structural correlation compared to purely structure-based models, underscoring the value of employing DMS data in predictive models.

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.

Building Machine Learning Small Molecule Melting Points and Solubility Models Using CCDC Melting Points Dataset

Predicting solubility of small molecules is a very difficult undertaking due to the lack of reliable and consistent experimental solubility data. It is well known that for a molecule in a crystal lattice to be dissolved, it must, first, dissociate from the lattice and then, second, be solvated. The melting point of a compound is proportional to the lattice energy, and the octanol-water partition coefficient (log P) is a measure of the compound's solvation efficiency. The CCDC's melting point dataset of almost one hundred thousand compounds was utilized to create widely applicable machine learning models of small molecule melting points. Using the general solubility equation, the aqueous thermodynamic solubilities of the same compounds can be predicted. The global model could be easily localized by adding additional melting point measurements for a chemical series of interest.

Dimensionally reduced machine learning model for predicting single component octanol-water partition coefficients

MF-LOGP, a new method for determining a single component octanol-water partition coefficients ([Formula: see text]) is presented which uses molecular formula as the only input. Octanol-water partition coefficients are useful in many applications, ranging from environmental fate and drug delivery. Currently, partition coefficients are either experimentally measured or predicted as a function of structural fragments, topological descriptors, or thermodynamic properties known or calculated from precise molecular structures. The MF-LOGP method presented here differs from classical methods as it does not require any structural information and uses molecular formula as the sole model input. MF-LOGP is therefore useful for situations in which the structure is unknown or where the use of a low dimensional, easily automatable, and computationally inexpensive calculations is required. MF-LOGP is a random forest algorithm that is trained and tested on 15,377 data points, using 10 features derived from the molecular formula to make [Formula: see text] predictions. Using an independent validation set of 2713 data points, MF-LOGP was found to have an average [Formula: see text] = 0.77 ± 0.007, [Formula: see text] = 0.52 ± 0.003, and [Formula: see text] = 0.83 ± 0.003. This performance fell within the spectrum of performances reported in the published literature for conventional higher dimensional models ([Formula: see text] = 0.42-1.54, [Formula: see text] = 0.09-1.07, and [Formula: see text] = 0.32-0.95). Compared with existing models, MF-LOGP requires a maximum of ten features and no structural information, thereby providing a practical and yet predictive tool. The development of MF-LOGP provides the groundwork for development of more physical prediction models leveraging big data analytical methods or complex multicomponent mixtures.

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.

Molecular partition coefficient from machine learning with polarization and entropy embedded atom-centered symmetry functions

Efficient prediction of the partition coefficient (log P) between polar and non-polar phases could shorten the cycle of drug and materials design. In this work, a descriptor, named 〈q - ACSFs〉_conf, is proposed to take the explicit polarization effects in the polar phase and the conformation ensemble of energetic and entropic significance in the non-polar phase into consideration. The polarization effects are involved by embedding the partial charge directly derived from force fields or quantum chemistry calculations into the atom-centered symmetry functions (ACSFs), together with the entropy effects, which are averaged according to the Boltzmann distribution of different conformations taken from the similarity matrix. The model was trained with high-dimensional neural networks (HDNNs) on a public dataset PhysProp (with 41 039 samples). Satisfactory log P prediction performance was achieved on three other datasets, namely, Martel (707 molecules), Star & Non-Star (266) and Huuskonen (1870). The present 〈q - ACSFs〉_conf model was also applicable to n-carboxylic acids with the number of carbons ranging from 2 to 14 and 54 kinds of organic solvent. It is easy to apply the present method to arbitrary sized systems and give a transferable atom-based partition coefficient.

MRlogP: Transfer Learning Enables Accurate logP Prediction Using Small Experimental Training Datasets

Small molecule lipophilicity is often included in generalized rules for medicinal chemistry. These rules aim to reduce time, effort, costs, and attrition rates in drug discovery, allowing the rejection or prioritization of compounds without the need for synthesis and testing. The availability of high quality, abundant training data for machine learning methods can be a major limiting factor in building effective property predictors. We utilize transfer learning techniques to get around this problem, first learning on a large amount of low accuracy predicted logP values before finally tuning our model using a small, accurate dataset of 244 druglike compounds to create MRlogP, a neural network-based predictor of logP capable of outperforming state of the art freely available logP prediction methods for druglike small molecules. MRlogP achieves an average root mean squared error of 0.988 and 0.715 against druglike molecules from Reaxys and PHYSPROP. We have made the trained neural network predictor and all associated code for descriptor generation freely available. In addition, MRlogP may be used online via a web interface.

Strategies to apply 3Rs in preclinical testing

Animal experimentation has been fundamental in biological and biomedical research. To guarantee the maximum quality, efficacy and/or safety of products intended for the use in humans in vivo testing is necessary; however, for over 60 years, alternative methods have been developed in response to the necessity to reduce the number of animals used in experimentation, to guarantee their welfare; resorting to animal models only when strictly necessary. The three Rs (Replacement, Reduction, and Refinement), seek to ensure the rational and respectful use of laboratory animals and maintain an adequate projection in terms of bioethical considerations. This article describes different approaches to apply 3Rs in preclinical experimentation for either research or regulatory purposes.

Simulating drug concentrations in PDMS microfluidic organ chips

Microfluidic organ-on-a-chip (Organ Chip) cell culture devices are often fabricated using polydimethylsiloxane (PDMS) because it is biocompatible, transparent, elastomeric, and oxygen permeable; however, hydrophobic small molecules can absorb to PDMS, which makes it challenging to predict drug responses. Here, we describe a combined simulation and experimental approach to predict the spatial and temporal concentration profile of a drug under continuous dosing in a PDMS Organ Chip containing two parallel channels separated by a porous membrane that is lined with cultured cells, without prior knowledge of its log P value. First, a three-dimensional finite element model of drug loss into the chip was developed that incorporates absorption, adsorption, convection, and diffusion, which simulates changes in drug levels over time and space as a function of potential PDMS diffusion coefficients and log P values. By then experimentally measuring the diffusivity of the compound in PDMS and determining its partition coefficient through mass spectrometric analysis of the drug concentration in the channel outflow, it is possible to estimate the effective log P range of the compound. The diffusion and partition coefficients were experimentally derived for the antimalarial drug and potential SARS-CoV-2 therapeutic, amodiaquine, and incorporated into the model to quantitatively estimate the drug-specific concentration profile over time measured in human lung airway chips lined with bronchial epithelium interfaced with pulmonary microvascular endothelium. The same strategy can be applied to any device geometry, surface treatment, or in vitro microfluidic model to simulate the spatial and temporal gradient of a drug in 3D without prior knowledge of the partition coefficient or the rate of diffusion in PDMS. Thus, this approach may expand the use of PDMS Organ Chip devices for various forms of drug testing.

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.

MLSolvA: solvation free energy prediction from pairwise atomistic interactions by machine learning

Recent advances in machine learning technologies and their applications have led to the development of diverse structure-property relationship models for crucial chemical properties. The solvation free energy is one of them. Here, we introduce a novel ML-based solvation model, which calculates the solvation energy from pairwise atomistic interactions. The novelty of the proposed model consists of a simple architecture: two encoding functions extract atomic feature vectors from the given chemical structure, while the inner product between the two atomistic feature vectors calculates their interactions. The results of 6239 experimental measurements achieve outstanding performance and transferability for enlarging training data owing to its solvent-non-specific nature. An analysis of the interaction map shows that our model has significant potential for producing group contributions on the solvation energy, which indicates that the model provides not only predictions of target properties but also more detailed physicochemical insights.

Multitask machine learning models for predicting lipophilicity (logP) in the SAMPL7 challenge

Accurate prediction of lipophilicity—logP—based on molecular structures is a well-established field. Predictions of logP are often used to drive forward drug discovery projects. Driven by the SAMPL7 challenge, in this manuscript we describe the steps that were taken to construct a novel machine learning model that can predict and generalize well. This model is based on the recently described Directed-Message Passing Neural Networks (D-MPNNs). Further enhancements included: both the inclusion of additional datasets from ChEMBL (RMSE improvement of 0.03), and the addition of helper tasks (RMSE improvement of 0.04). To the best of our knowledge, the concept of adding predictions from other models (Simulations Plus logP and logD@pH7.4, respectively) as helper tasks is novel and could be applied in a broader context. The final model that we constructed and used to participate in the challenge ranked 2/17 ranked submissions with an RMSE of 0.66, and an MAE of 0.48 (submission: Chemprop). On other datasets the model also works well, especially retrospectively applied to the SAMPL6 challenge where it would have ranked number one out of all submissions (RMSE of 0.35). Despite the fact that our model works well, we conclude with suggestions that are expected to improve the model even further.

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.

ClassicalGSG: Prediction of log P using classical molecular force fields and geometric scattering for graphs

This work examines methods for predicting the partition coefficient (log P) for a dataset of small molecules. Here, we use atomic attributes such as radius and partial charge, which are typically used as force field parameters in classical molecular dynamics simulations. These atomic attributes are transformed into index-invariant molecular features using a recently developed method called geometric scattering for graphs (GSG). We call this approach "ClassicalGSG" and examine its performance under a broad range of conditions and hyperparameters. We train ClassicalGSG log P predictors with neural networks using 10,722 molecules from the OpenChem dataset and apply them to predict the log P values from four independent test sets. The ClassicalGSG method's performance is compared to a baseline model that employs graph convolutional networks. Our results show that the best prediction accuracies are obtained using atomic attributes generated with the CHARMM generalized force field and 2D molecular structures.

Computational Identification of Novel Families of Nonfullerene Acceptors by Modification of Known Compounds

We considered a database of tens of thousands of known organic semiconductors and identified those compounds with computed electronic properties (orbital energies, excited state energies, and oscillator strengths) that would make them suitable as nonfullerene electron acceptors in organic solar cells. The range of parameters for the desirable acceptors is determined from a set of experimentally characterized high-efficiency nonfullerene acceptors. This search leads to ∼30 lead compounds never considered before for organic photovoltaic applications. We then proceed to modify these compounds to bring their computed solubility in line with that of the best small-molecule nonfullerene acceptors. A further refinement of the search can be based on additional properties like the reorganization energy for chemical reduction. This simple strategy, which relies on a few easily computable parameters and can be expanded to a larger set of molecules, enables the identification of completely new chemical families to be explored experimentally.

Predicting the membrane permeability of organic fluorescent probes by the deep neural network based lipophilicity descriptor DeepFl-LogP

Light microscopy has become an indispensable tool for the life sciences, as it enables the rapid acquisition of three-dimensional images from the interior of living cells/tissues. Over the last decades, super-resolution light microscopy techniques have been developed, which allow a resolution up to an order of magnitude higher than that of conventional light microscopy. Those techniques require labelling of cellular structures with fluorescent probes exhibiting specific properties, which are supplied from outside and therefore have to surpass cell membranes. Currently, major efforts are undertaken to develop probes which can surpass cell membranes and exhibit the photophysical properties required for super-resolution imaging. However, the process of probe development is still based on a tedious and time consuming manual screening. An accurate computer based model that enables the prediction of the cell permeability based on their chemical structure would therefore be an invaluable asset for the development of fluorescent probes. Unfortunately, current models, which are based on multiple molecular descriptors, are not well suited for this task as they require high effort in the usage and exhibit moderate accuracy in their prediction. Here, we present a novel fragment based lipophilicity descriptor DeepFL-LogP, which was developed on the basis of a deep neural network. DeepFL-LogP exhibits excellent correlation with the experimental partition coefficient reference data (R2 = 0.892 and MSE = 0.359) of drug-like substances. Further a simple threshold permeability model on the basis of this descriptor allows to categorize the permeability of fluorescent probes with 96% accuracy. This novel descriptor is expected to largely simplify and speed up the development process for novel cell permeable fluorophores.

Enhancing Carbon Acid pKa Prediction by Augmentation of Sparse Experimental Datasets with Accurate AIBL (QM) Derived Values

The prediction of the aqueous pKa of carbon acids by Quantitative Structure Property Relationship or cheminformatics-based methods is a rather arduous problem. Primarily, there are insufficient high-quality experimental data points measured in homogeneous conditions to allow for a good global model to be generated. In our computationally efficient pKa prediction method, we generate an atom-type feature vector, called a distance spectrum, from the assigned ionisation atom, and learn coefficients for those atom-types that show the impact each atom-type has on the pKa of the ionisable centre. In the current work, we augment our dataset with pKa values from a series of high performing local models derived from the Ab Initio Bond Lengths method (AIBL). We find that, in distilling the knowledge available from multiple models into one general model, the prediction error for an external test set is reduced compared to that using literature experimental data alone.

Machine Learning Predicts Degree of Aromaticity from Structural Fingerprints

Prediction of whether a compound is "aromatic" is at first glance a relatively simple task-does it obey Hückel's rule (planar cyclic π-system with 4n + 2 electrons) or not? However, aromaticity is far from a binary property, and there are distinct variations in the chemical and biological behavior of different systems which obey Hückel's rule and are thus classified as aromatic. To that end, the aromaticity of each molecule in a large public dataset was quantified by an extension of the work of Raczyńska et al. Building on this data, a method is proposed for machine learning the degree of aromaticity of each aromatic ring in a molecule. Categories are derived from the numeric results, allowing the differentiation of structural patterns between them and thus a better representation of the underlying chemical and biological behavior in expert and (Q)SAR systems.

Network Analysis for Prioritizing Biodegradation Metabolites of Polycyclic Aromatic Hydrocarbons

Polycyclic aromatic hydrocarbons (PAHs) are a diverse group of environmental contaminants released during the combustion of organic materials and the production and utilization of fossil fuels. Once released, PAHs deposit in soil and water bodies where they are subjected to environmental transport and transformations. As they degrade, intermediate transformation products may play an important role in their environmental impact. However, studying the effects of these degradation products has proven challenging because of the complexity, transience, and low concentration of many intermediates. Herein, a novel integration of a pathway prediction system and network theory was developed and applied to a set of four PAHs to demonstrate a possible solution to this challenge. Network analysis techniques were employed to refine the thousands of potential outputs and elucidate compounds of interest. Using these tools, we determined correlations between PAH degradation network data and intermediate metabolite structures, gaining information about the chemical characteristics of compounds based on their placement within the degradation network. Upon applying our developed filtering algorithm, we are able to predict up to 48% of the most common transformation products identified in a comprehensive empirical literature review. Additionally, our integrated approach uncovers potential metabolites which connect those found by past empirical studies but are currently undetected, thereby filling in the gaps of information in PAH degradation pathways.

A deep learning approach for the blind logP prediction in SAMPL6 challenge

Water octanol partition coefficient serves as a measure for the lipophilicity of a molecule and is important in the field of drug discovery. A novel method for computational prediction of logarithm of partition coefficient (logP) has been developed using molecular fingerprints and a deep neural network. The machine learning model was trained on a dataset of 12,000 molecules and tested on 2000 molecules. In this article, we present our results for the blind prediction of logP for the SAMPL6 challenge. While the best submission achieved a RMSE of 0.41 logP units, our submission had a RMSE of 0.61 logP units. Overall, we ranked in the top quarter out of the 92 submissions that were made. Our results show that the deep learning model can be used as a fast, accurate and robust method for high throughput prediction of logP of small molecules.

Global Sensitivity Analysis of the Rodgers and Rowland Model for Prediction of Tissue: Plasma Partitioning Coefficients: Assessment of the Key Physiological and Physicochemical Factors That Determine Small-Molecule Tissue Distribution

In physiologically based pharmacokinetic (PBPK) modelling, the large number of input parameters, limited amount of available data and the structural model complexity generally hinder simultaneous estimation of uncertain and/or unknown parameters. These parameters are generally subject to estimation. However, the approaches taken for parameter estimation vary widely. Global sensitivity analyses are proposed as a method to systematically determine the most influential parameters that can be subject to estimation. Herein, a global sensitivity analysis was conducted to identify the key drug and physiological parameters influencing drug disposition in PBPK models and to potentially reduce the PBPK model dimensionality. The impact of these parameters was evaluated on the tissue-to-unbound plasma partition coefficients (Kpus) predicted by the Rodgers and Rowland model using Latin hypercube sampling combined to partial rank correlation coefficients (PRCC). For most drug classes, PRCC showed that LogP and fraction unbound in plasma (fu_p) were generally the most influential parameters for Kpu predictions. For strong bases, blood:plasma partitioning was one of the most influential parameter. Uncertainty in tissue composition parameters had a large impact on Kpu and Vss predictions for all classes. Among tissue composition parameters, changes in Kpu outputs were especially attributed to changes in tissue acidic phospholipid concentrations and extracellular protein tissue:plasma ratio values. In conclusion, this work demonstrates that for parameter estimation involving PBPK models and dimensionality reduction purposes, less influential parameters might be assigned fixed values depending on the parameter space, while influential parameters could be subject to parameters estimation.

A comparison of molecular representations for lipophilicity quantitative structure-property relationships with results from the SAMPL6 logP Prediction Challenge

Effective representation of a molecule is required to develop useful quantitative structure-property relationships (QSPR) for accurate prediction of chemical properties. The octanol-water partition coefficient logP, a measure of lipophilicity, is an important property for pharmacological and toxicological endpoints used in the pharmaceutical and regulatory spheres. We compare physicochemical descriptors, structural keys, and circular fingerprints in their ability to effectively represent a chemical space and characterise molecular features to correlate with lipophilicity. Exploratory landscape continuity analyses revealed that whole-molecule physicochemical descriptors could map together compounds that were similar in both molecular features and logP, indicating higher potential for use in logP QSPRs compared to the substructural approach of structural keys and circular fingerprints. Indeed, logP QSPR models parameterised by physicochemical descriptors consistently performed with the lowest error. Our best performing model was a stochastic gradient descent-optimised multilinear regression with 1438 descriptors, returning an internal benchmark RMSE of 1.03 log units. This corroborates the well-established notion that lipophilicity is an additive, whole-molecule property. We externally tested the model by participating in the 2019 SAMPL6 logP Prediction Challenge and blindly predicting for 11 protein kinase inhibitor fragment-like molecules. Our model returned an RMSE of 0.49 log units, placing eighth overall and third in the empirical methods category (submission ID 'hdpuj'). Permutation feature importance analyses revealed that physicochemical descriptors could characterise predictive molecular features highly relevant to the kinase inhibitor fragment-like molecules.

Avoiding hERG-liability in drug design via synergetic combinations of different (Q)SAR methodologies and data sources: a case study in an industrial setting

In this paper, we explore the impact of combining different in silico prediction approaches and data sources on the predictive performance of the resulting system. We use inhibition of the hERG ion channel target as the endpoint for this study as it constitutes a key safety concern in drug development and a potential cause of attrition. We will show that combining data sources can improve the relevance of the training set in regard of the target chemical space, leading to improved performance. Similarly we will demonstrate that combining multiple statistical models together, and with expert systems, can lead to positive synergistic effects when taking into account the confidence in the predictions of the merged systems. The best combinations analyzed display a good hERG predictivity. Finally, this work demonstrates the suitability of the SOHN methodology for building models in the context of receptor based endpoints like hERG inhibition when using the appropriate pharmacophoric descriptors.