Systematic Approaches for the Encoding of Chemical Groups: A Case Study.

Regulatory authorities aim to organize substances into groups to facilitate prioritization within hazard and risk assessment processes. Often, such chemical groupings are not explicitly defined by structural rules or physicochemical property information. This is largely due to how these groupings are developed, namely, a manual expert curation process, which in turn makes updating and refining groupings, as new substances are evaluated, a practical challenge. Herein, machine learning methods were leveraged to build models that could preliminarily assign substances to predefined groups. A set of 86 groupings containing 2,184 substances as published on the European Chemicals Agency (ECHA) website were mapped to the U.S. Environmental Protection Agency (EPA) Distributed Toxicity Structure Database (DSSTox) content to extract chemical and structural information. Substances were represented using Morgan fingerprints, and two machine learning approaches were used to classify test substances into 56 groups containing at least 10 substances with a structural representation in the data set: k-nearest neighbor (kNN) and random forest (RF), that led to mean 5-fold cross-validation test accuracies (average F1 scores) of 0.781 and 0.853, respectively. With a 9% improvement, the RF classifier was significantly more accurate than KNN (p-value = 0.001). The approach offers promise as a means of the initial profiling of new substances into predefined groups to facilitate prioritization efforts and streamline the assessment of new substances when earlier groupings are available. The algorithm to fit and use these models has been made available in the accompanying repository, thereby enabling both use of the produced models and refitting of these models, as new groupings become available by regulatory authorities or industry.

A non-clinical and clinical IUCLID database for 530 pharmaceuticals (part I): Methodological aspects of its development.

A new IUCLID database is provided containing results from non-clinical animal studies and human information for 530 approved drugs. The database was developed by extracting data from pharmacological reviews of repeat-dose, carcinogenicity, developmental, and reproductive toxicity studies. In the database, observed and no-observed effects are linked to the respective effect levels, including information on severity/incidence and transiency/reversibility. It also includes some information on effects in humans, that were extracted from relevant sections of standard product labels of the approved drugs. The database is complemented with a specific ontology for reporting effects that was developed as an improved version of the Ontology Lookup Service's mammalian and human phenotype ontologies and includes different hierarchical levels. The developed ontology contains novel and unique standardized terms, including ontological terms for reproductive and endocrine effects. The database aims to facilitate correlation and concordance analyses based on the link between observed and no-observed effects and their respective effect levels. In addition, it offers a robust dataset on drug information for the pharmaceutical industry and research. The reported ontology supports the analyses of toxicological information, especially for reproductive and endocrine endpoints and can be used to encode legacy data or develop additional ontologies. The new database and ontology can be used to support the development of alternative non-animal approaches, to elucidate mechanisms of toxicity, and to analyse human relevance. The new IUCLID database is provided free of charge at https://iuclid6.echa.europa.eu/us-fda-toxicity-data.

Utility of In Vitro Bioactivity as a Lower Bound Estimate of In Vivo Adverse Effect Levels and in Risk-Based Prioritization.

Use of high-throughput, in vitro bioactivity data in setting a point-of-departure (POD) has the potential to accelerate the pace of human health safety evaluation by informing screening-level assessments. The primary objective of this work was to compare PODs based on high-throughput predictions of bioactivity, exposure predictions, and traditional hazard information for 448 chemicals. PODs derived from new approach methodologies (NAMs) were obtained for this comparison using the 50th (PODNAM, 50) and the 95th (PODNAM, 95) percentile credible interval estimates for the steady-state plasma concentration used in in vitro to in vivo extrapolation of administered equivalent doses. Of the 448 substances, 89% had a PODNAM, 95 that was less than the traditional POD (PODtraditional) value. For the 48 substances for which PODtraditional < PODNAM, 95, the PODNAM and PODtraditional were typically within a factor of 10 of each other, and there was an enrichment of chemical structural features associated with organophosphate and carbamate insecticides. When PODtraditional < PODNAM, 95, it did not appear to result from an enrichment of PODtraditional based on a particular study type (eg, developmental, reproductive, and chronic studies). Bioactivity:exposure ratios, useful for identification of substances with potential priority, demonstrated that high-throughput exposure predictions were greater than the PODNAM, 95 for 11 substances. When compared with threshold of toxicological concern (TTC) values, the PODNAM, 95 was greater than the corresponding TTC value 90% of the time. This work demonstrates the feasibility, and continuing challenges, of using in vitro bioactivity as a protective estimate of POD in screening-level assessments via a case study.

The impact on classifications for carcinogenicity, mutagenicity, reproductive and specific target organ toxicity after repeated exposure in the first ten years of the REACH regulation.

The present study primarily aims at informing regulators and policy makers in Europe and examines the evolution of self-classifications and study availability for the endpoints of carcinogenicity, mutagenicity, reproductive toxicity (CMR) and specific target organ toxicity after repeated exposure (STOT RE) for the first ten years of the REACH legislation. Our knowledge on chemical safety keeps increasing due to the registration obligations under REACH, in combination with proactive actions by registrants and regulatory actions by Authorities, which jointly lead to new testing and critical reassessment of existing studies. The improvements become evident by the constant increase in the number of substances that are self-classified by the registrants for human health endpoints. Moreover, there is a slow but steady increase in the number of substances for which there is at least one experimental study available for the human health endpoints in scope of this analysis. However, the increase is slow given the generally limited data availability at the beginning of REACH. Manual examination of about 350 classified substances reveals that the impact of newly generated data and regulatory action by Authorities is greater for reproductive toxicity than for carcinogenicity or mutagenicity, reflecting the strengthening of the information requirements for reproductive toxicity with the introduction of REACH. The results of the study should inform regulators and policy makers at EU and national level in the discussion on potential changes to information requirements or testing strategies under REACH.

(Q)SARs as Adaptations to REACH Information Requirements.

REACH is a regulation of the European Union adopted to improve the safe use of chemicals with regard to human health and the environment. The safe use of chemicals can be achieved only if the hazard and the exposure of the substances are well characterized. Testing on animals has been traditionally the main tool for hazard assessment. For ethical and economic reasons, alternative ways of testing that do not use laboratory animals have been developed by different parties (regulatory agencies, researchers, industry) over the recent decades, and their proper use in hazard assessment is encouraged under REACH. In this chapter, we describe how (Q)SAR models and predictions are included into REACH and their adequate use promoted by the European Chemicals Agency (ECHA).

UVCB substances: methodology for structural description and application to fate and hazard assessment.

Substances of unknown or variable composition, complex reaction products, or biological materials (UVCBs) have been conventionally described in generic terms. Commonly used substance identifiers are generic names of chemical classes, generic structural formulas, reaction steps, physical-chemical properties, or spectral data. Lack of well-defined structural information has significantly restricted in silico fate and hazard assessment of UVCB substances. A methodology for the structural description of UVCB substances has been developed that allows use of known identifiers for coding, generation, and selection of representative constituents. The developed formats, Generic Simplified Molecular-Input Line-Entry System (G SMILES) and Generic Graph (G Graph), address the need to code, generate, and select representative UVCB constituents; G SMILES is a SMILES-based single line notation coding fixed and variable structural features of UVCBs, whereas G Graph is based on a workflow paradigm that allows generation of constituents coded in G SMILES and end point-specific or nonspecific selection of representative constituents. Structural description of UVCB substances as afforded by the developed methodology is essential for in silico fate and hazard assessment. Data gap filling approaches such as read-across, trend analysis, or quantitative structure-activity relationship modeling can be applied to the generated constituents, and the results can be used to assess the substance as a whole. The methodology also advances the application of category-based data gap filling approaches to UVCB substances.

Computer-aided molecular design of solvents for accelerated reaction kinetics.

Solvents can significantly alter the rates and selectivity of liquid-phase organic reactions, often hindering the development of new synthetic routes or, if chosen wisely, facilitating routes by improving rates and selectivities. To address this challenge, a systematic methodology is proposed that quickly identifies improved reaction solvents by combining quantum mechanical computations of the reaction rate constant in a few solvents with a computer-aided molecular design (CAMD) procedure. The approach allows the identification of a high-performance solvent within a very large set of possible molecules. The validity of our CAMD approach is demonstrated through application to a classical nucleophilic substitution reaction for the study of solvent effects, the Menschutkin reaction. The results were validated successfully by in situ kinetic experiments. A space of 1,341 solvents was explored in silico, but required quantum-mechanical calculations of the rate constant in only nine solvents, and uncovered a solvent that increases the rate constant by 40%.

The polymorphs of ROY: application of a systematic crystal structure prediction technique.

We investigate the ability of current ab initio crystal structure prediction techniques to identify the polymorphs of 5-methyl-2-[(2-nitrophenyl)amino]-3-thiophenecarbonitrile, also known as ROY because of the red, orange and yellow colours of its polymorphs. We use a methodology combining the generation of a large number of structures based on a computationally inexpensive model using the CrystalPredictor global search algorithm, and the further minimization of the most promising of these structures using the CrystalOptimizer local minimization algorithm which employs an accurate, yet efficiently constructed, model based on isolated-molecule quantum-mechanical calculations. We demonstrate that this approach successfully predicts the seven experimentally resolved structures of ROY as lattice-energy minima, with five of these structures being within the 12 lowest energy structures predicted. Some of the other low-energy structures identified are likely candidates for the still unresolved polymorphs of this molecule. The relative stability of the predicted structures only partially matches that of the experimentally resolved polymorphs. The worst case is that of polymorph ON, whose relative energy with respect to Y is overestimated by 6.65 kJ mol(-1). This highlights the need for further developments in the accuracy of the energy calculations.

Towards crystal structure prediction of complex organic compounds--a report on the fifth blind test.

Following on from the success of the previous crystal structure prediction blind tests (CSP1999, CSP2001, CSP2004 and CSP2007), a fifth such collaborative project (CSP2010) was organized at the Cambridge Crystallographic Data Centre. A range of methodologies was used by the participating groups in order to evaluate the ability of the current computational methods to predict the crystal structures of the six organic molecules chosen as targets for this blind test. The first four targets, two rigid molecules, one semi-flexible molecule and a 1:1 salt, matched the criteria for the targets from CSP2007, while the last two targets belonged to two new challenging categories - a larger, much more flexible molecule and a hydrate with more than one polymorph. Each group submitted three predictions for each target it attempted. There was at least one successful prediction for each target, and two groups were able to successfully predict the structure of the large flexible molecule as their first place submission. The results show that while not as many groups successfully predicted the structures of the three smallest molecules as in CSP2007, there is now evidence that methodologies such as dispersion-corrected density functional theory (DFT-D) are able to reliably do so. The results also highlight the many challenges posed by more complex systems and show that there are still issues to be overcome.

Successful prediction of a model pharmaceutical in the fifth blind test of crystal structure prediction.

The range of target structures in the fifth international blind test of crystal structure prediction was extended to include a highly flexible molecule, (benzyl-(4-(4-methyl-5-(p-tolylsulfonyl)-1,3-thiazol-2-yl)phenyl)carbamate, as a challenge representative of modern pharmaceuticals. Two of the groups participating in the blind test independently predicted the correct structure. The methods they used are described and contrasted, and the implications of the capability to tackle molecules of this complexity are discussed.

Which, if any, hydrates will crystallise? Predicting hydrate formation of two dihydroxybenzoic acids.

A study of two dihydroxybenzoic acid isomers shows that computational methods can be used to predict hydrate formation, the compound:water ratio and hydrate crystal structures. The calculations also help identify a novel hydrate found in the solid form screening that validates this study.

Solid-State Forms of ß-Resorcylic Acid: How Exhaustive Should a Polymorph Screen Be?

An extensive experimental screen, coupled with a computational study, revealed seven new solid-state forms of ß-resorcylic acid. The known, stable polymorph II° shows a reversible phase transformation to the new, kinetically stable, probably disordered high temperature form I. The study provides a consistent picture of the solid-state of ß-resorcylic acid.

Modelling organic crystal structures using distributed multipole and polarizability-based model intermolecular potentials.

Crystal structure prediction for organic molecules requires both the fast assessment of thousands to millions of crystal structures and the greatest possible accuracy in their relative energies. We describe a crystal lattice simulation program, DMACRYS, emphasizing the features that make it suitable for use in crystal structure prediction for pharmaceutical molecules using accurate anisotropic atom-atom model intermolecular potentials based on the theory of intermolecular forces. DMACRYS can optimize the lattice energy of a crystal, calculate the second derivative properties, and reduce the symmetry of the spacegroup to move away from a transition state. The calculated terahertz frequency k = 0 rigid-body lattice modes and elastic tensor can be used to estimate free energies. The program uses a distributed multipole electrostatic model (Q, t = 00,...,44s) for the electrostatic fields, and can use anisotropic atom-atom repulsion models, damped isotropic dispersion up to R(-10), as well as a range of empirically fitted isotropic exp-6 atom-atom models with different definitions of atomic types. A new feature is that an accurate model for the induction energy contribution to the lattice energy has been implemented that uses atomic anisotropic dipole polarizability models (alpha, t = (10,10)...(11c,11s)) to evaluate the changes in the molecular charge density induced by the electrostatic field within the crystal. It is demonstrated, using the four polymorphs of the pharmaceutical carbamazepine C(15)H(12)N(2)O, that whilst reproducing crystal structures is relatively easy, calculating the polymorphic energy differences to the accuracy of a few kJ mol(-1) required for applications is very demanding of assumptions made in the modelling. Thus DMACRYS enables the comparison of both known and hypothetical crystal structures as an aid to the development of pharmaceuticals and other speciality organic materials, and provides a tool to develop the modelling of the intermolecular forces involved in molecular recognition processes.

The Use of Anisotropic Potentials in Modeling Water and Free Energies of Hydration.

We propose a novel, anisotropic rigid-body intermolecular potential model to predict the properties of water and the hydration free energies of neutral organic solutes. The electrostatic interactions of water and the solutes are modeled using atomic multipole moments up to hexadecapole; these are obtained from distributed multipole analysis of the quantum mechanically computed charge densities and include average polarization effects in solution. The repulsion-dispersion water-water interactions are modeled with a three-site, exp-6 model fitted to the experimental liquid water density and oxygen-oxygen radial distribution function at ambient conditions. The proposed water model reproduces well several water properties not used in its parametrization, including vapor-liquid coexistence densities, the maximum in liquid water density at atmospheric pressure, the structure of ordered ice polymorphs, and the liquid water heat capacity. The model is used to compute the hydration free energy of 10 neutral organic solutes using explicit-solvent free energy perturbation. The solute-solute repulsion-dispersion intermolecular potential is obtained from previous parametrizations on organic crystal structures. In order to calculate the free energies of hydration, water-solute repulsion-dispersion interactions are modeled using Lorenz-Berthelot combining rules. The root-mean-square error of the predicted hydration free energies is 1.5 kcal mol(-1), which is comparable to the error found using a continuum mean-field quantum mechanical approach parametrized using experimental free energy of hydration data. The results are also contrasted with explicit-solvent hydration free energies obtained with an atomic charge representation of the solute's charge density computed at the same level of theory used to compute the distributed multipoles. Replacing the multipole description of the solute's charge density with an atomic charge model changes the free energy of hydration by as much as 3 kcal mol(-1) and provides an estimate for the effect of the modeling quality of the intermolecular electrostatic forces in free energy of solvation calculations.

A computationally inspired investigation of the solid forms of (R)-1-phenylethylammonium-(S)-2-phenylbutyrate.

Following the computation of a lattice energy landscape which predicted that there should be more stable, denser forms of (R)-1-phenylethylammonium-(S)-2-phenylbutyrate, crystallizations from a range of solvents were performed to search for other polymorphs and investigate the possibility that the known P4(1) structure could be a hydrate. Extensive crystallization experiments from a wide range of solvents gave fine needles or microcrystalline samples. A redetermination of the P4(1) structure by powder X-ray diffraction located all protons, and in conjunction with other experimental and computational evidence showed that the structure was anhydrous. Evidence for two additional forms was found as mixtures with form I. These include an orthorhombic form, possibly a Z' = 3 polymorph, and another as yet unidentified form obtained as a minor component from dichloromethane solution. However, both these forms appear to be metastable with respect to form I (P4(1)), which is therefore probably the most thermodynamically stable form that can be crystallized from solution under ambient conditions. This determination of the solid state behavior of the less readily crystallized member of the diastereomeric salt system (R)-1-phenylethylammonium-(R/S)-2-phenylbutyrate provides a challenge to the theoretical modeling to explain its ideal resolution behavior.

Significant progress in predicting the crystal structures of small organic molecules--a report on the fourth blind test.

We report on the organization and outcome of the fourth blind test of crystal structure prediction, an international collaborative project organized to evaluate the present state in computational methods of predicting the crystal structures of small organic molecules. There were 14 research groups which took part, using a variety of methods to generate and rank the most likely crystal structures for four target systems: three single-component crystal structures and a 1:1 cocrystal. Participants were challenged to predict the crystal structures of the four systems, given only their molecular diagrams, while the recently determined but as-yet unpublished crystal structures were withheld by an independent referee. Three predictions were allowed for each system. The results demonstrate a dramatic improvement in rates of success over previous blind tests; in total, there were 13 successful predictions and, for each of the four targets, at least two groups correctly predicted the observed crystal structure. The successes include one participating group who correctly predicted all four crystal structures as their first ranked choice, albeit at a considerable computational expense. The results reflect important improvements in modelling methods and suggest that, at least for the small and fairly rigid types of molecules included in this blind test, such calculations can be constructively applied to help understand crystallization and polymorphism of organic molecules.

Can the Formation of Pharmaceutical Cocrystals Be Computationally Predicted? 2. Crystal Structure Prediction.

We report a multistage lattice energy minimization methodology for generating stable packing arrangements of cocrystals containing flexible molecules. In the first approximation, the intermolecular electrostatic interactions are modeled with atomic charges and the molecular deformation energy is interpolated over a set of precomputed quantum mechanical values. At subsequent stages, the accuracy is improved by first using analytically rotated and then conformation-dependent multipole moments, computed from the isolated-molecule charge density, and "on-the-fly" quantum mechanical calculations to compute the intramolecular deformation energy. This multistage approach increases the efficiency of the search and establishes the molecule-dependent error due to the atomic charge representation of the charge density and the neglect of the conformational dependence of atomic multipole moments. The methodology is used to study the lattice energy landscapes of the cocrystals of 4-aminobenzoic acid with 2,2'-bipyridine and 4-nitrophenylacetic acid, as well as the single-component crystals. All single-component, experimentally determined crystal structures within the scope of the search were found at, or very close to, the global minimum. The experimental cocrystal with 2,2'-bipyridine is also predicted to be among the most stable packing arrangements. On the contrary, the lattice energy landscape of the cocrystal with 4-nitrophenylacetic acid contains several low energy structures that are more stable than the experimentally observed form and have different hydrogen bonding motifs. Overall, the methodology can provide worthwhile crystal energy landscapes for multicomponent organic solids and thereby contribute to understanding cocrystal formation.

Modeling the interplay of inter- and intramolecular hydrogen bonding in conformational polymorphs.

The predicted stability differences of the conformational polymorphs of oxalyl dihydrazide and ortho-acetamidobenzamide are unrealistically large when the modeling of intermolecular energies is solely based on the isolated-molecule charge density, neglecting charge density polarization. Ab initio calculated crystal electron densities showed qualitative differences depending on the spatial arrangement of molecules in the lattice with the greatest variations observed for polymorphs that differ in the extent of inter- and intramolecular hydrogen bonding. We show that accounting for induction dramatically alters the calculated stability order of the polymorphs and reduces their predicted stability differences to be in better agreement with experiment. Given the challenges in modeling conformational polymorphs with marked differences in hydrogen bonding geometries, we performed an extensive periodic density functional study with a range of exchange-correlation functionals using both atomic and plane wave basis sets. Although such electronic structure methods model the electrostatic and polarization contributions well, the underestimation of dispersion interactions by current exchange-correlation functionals limits their applicability. The use of an empirical dispersion-corrected density functional method consistently reduces the structural deviations between the experimental and energy minimized crystal structures and achieves plausible stability differences. Thus, we have established which types of models may give worthwhile relative energies for crystal structures and other condensed phases of flexible molecules with intra- and intermolecular hydrogen bonding capabilities, advancing the possibility of simulation studies on polymorphic pharmaceuticals.

The thermal stability of lattice-energy minima of 5-fluorouracil: metadynamics as an aid to polymorph prediction.

This paper reports a novel methodology for the free-energy minimization of crystal structures exhibiting strong, anisotropic interactions due to hydrogen bonding. The geometry of the thermally expanded cell was calculated by exploiting the dependence of the free-energy derivatives with respect to cell lengths and angles on the average pressure tensor computed in short molecular dynamics simulations. All dynamic simulations were performed with an elaborate anisotropic potential based on a distributed multipole analysis of the isolated molecule charge density. Changes in structure were monitored via simulated X-ray diffraction patterns. The methodology was used to minimize the free energy at ambient conditions of a set of experimental and hypothetical 5-fluorouracil crystal structures, generated in a search for lattice-energy minima with the same model potential. Our results demonstrate that the majority ( approximately 75%) of lattice-energy minima are thermally stable at ambient conditions, and hence, the free-energy (like the lattice-energy) surface is complex and highly undulating. Metadynamics trajectories (Laio, A.; Parrinello, M. Proc. Natl. Acad. Sci. U.S.A. 2002, 99, 12562) started from the free-energy minima only produced transitions that preserved the hydrogen-bonding motif, and thus, further developments are needed for this method to efficiently explore such free-energy surfaces. The existence of so many free-energy minima, with large barriers for the alteration of the hydrogen-bonding motif, is consistent with the range of motifs observed in crystal structures of 5-fluorouracil and other 5-substituted uracils.