Practical Aspects of Designing and Conducting Validation Studies Involving Multi-study Trials.

This chapter focuses on practical aspects of conducting prospective in vitro validation studies, and in particular, by laboratories that are members of the European Union Network of Laboratories for the Validation of Alternative Methods (EU-NETVAL) that is coordinated by the EU Reference Laboratory for Alternatives to Animal Testing (EURL ECVAM). Prospective validation studies involving EU-NETVAL, comprising a multi-study trial involving several laboratories or "test facilities", typically consist of two main steps: (1) the design of the validation study by EURL ECVAM and (2) the execution of the multi-study trial by a number of qualified laboratories within EU-NETVAL, coordinated and supported by EURL ECVAM. The approach adopted in the conduct of these validation studies adheres to the principles described in the OECD Guidance Document on the Validation and International Acceptance of new or updated test methods for Hazard Assessment No. 34 (OECD 2005). The context and scope of conducting prospective in vitro validation studies is dealt with in Chap. 4 . Here we focus mainly on the processes followed to carry out a prospective validation of in vitro methods involving different laboratories with the ultimate aim of generating a dataset that can support a decision in relation to the possible development of an international test guideline (e.g. by the OECD) or the establishment of performance standards.

In Silico Models for Acute Systemic Toxicity.

In this chapter, we give an overview of the regulatory requirements for acute systemic toxicity information in the European Union, and we review the availability of structure-based computational models that are available and potentially useful in the assessment of acute systemic toxicity. The most recently published literature models for acute systemic toxicity are also discussed, and perspectives for future developments in this field are offered.

CERAPP: Collaborative Estrogen Receptor Activity Prediction Project.

BACKGROUND: Humans are exposed to thousands of man-made chemicals in the environment. Some chemicals mimic natural endocrine hormones and, thus, have the potential to be endocrine disruptors. Most of these chemicals have never been tested for their ability to interact with the estrogen receptor (ER). Risk assessors need tools to prioritize chemicals for evaluation in costly in vivo tests, for instance, within the U.S. EPA Endocrine Disruptor Screening Program. OBJECTIVES: We describe a large-scale modeling project called CERAPP (Collaborative Estrogen Receptor Activity Prediction Project) and demonstrate the efficacy of using predictive computational models trained on high-throughput screening data to evaluate thousands of chemicals for ER-related activity and prioritize them for further testing. METHODS: CERAPP combined multiple models developed in collaboration with 17 groups in the United States and Europe to predict ER activity of a common set of 32,464 chemical structures. Quantitative structure-activity relationship models and docking approaches were employed, mostly using a common training set of 1,677 chemical structures provided by the U.S. EPA, to build a total of 40 categorical and 8 continuous models for binding, agonist, and antagonist ER activity. All predictions were evaluated on a set of 7,522 chemicals curated from the literature. To overcome the limitations of single models, a consensus was built by weighting models on scores based on their evaluated accuracies. RESULTS: Individual model scores ranged from 0.69 to 0.85, showing high prediction reliabilities. Out of the 32,464 chemicals, the consensus model predicted 4,001 chemicals (12.3%) as high priority actives and 6,742 potential actives (20.8%) to be considered for further testing. CONCLUSION: This project demonstrated the possibility to screen large libraries of chemicals using a consensus of different in silico approaches. This concept will be applied in future projects related to other end points. CITATION: Mansouri K, Abdelaziz A, Rybacka A, Roncaglioni A, Tropsha A, Varnek A, Zakharov A, Worth A, Richard AM, Grulke CM, Trisciuzzi D, Fourches D, Horvath D, Benfenati E, Muratov E, Wedebye EB, Grisoni F, Mangiatordi GF, Incisivo GM, Hong H, Ng HW, Tetko IV, Balabin I, Kancherla J, Shen J, Burton J, Nicklaus M, Cassotti M, Nikolov NG, Nicolotti O, Andersson PL, Zang Q, Politi R, Beger RD, Todeschini R, Huang R, Farag S, Rosenberg SA, Slavov S, Hu X, Judson RS. 2016. CERAPP: Collaborative Estrogen Receptor Activity Prediction Project. Environ Health Perspect 124:1023-1033; http://dx.doi.org/10.1289/ehp.1510267.

Chemical Safety Assessment Using Read-Across: Assessing the Use of Novel Testing Methods to Strengthen the Evidence Base for Decision Making.

BACKGROUND: Safety assessment for repeated dose toxicity is one of the largest challenges in the process to replace animal testing. This is also one of the proof of concept ambitions of SEURAT-1, the largest ever European Union research initiative on alternative testing, co-funded by the European Commission and Cosmetics Europe. This review is based on the discussion and outcome of a workshop organized on initiative of the SEURAT-1 consortium joined by a group of international experts with complementary knowledge to further develop traditional read-across and include new approach data. OBJECTIVES: The aim of the suggested strategy for chemical read-across is to show how a traditional read-across based on structural similarities between source and target substance can be strengthened with additional evidence from new approach data--for example, information from in vitro molecular screening, "-omics" assays and computational models--to reach regulatory acceptance. METHODS: We identified four read-across scenarios that cover typical human health assessment situations. For each such decision context, we suggested several chemical groups as examples to prove when read-across between group members is possible, considering both chemical and biological similarities. CONCLUSIONS: We agreed to carry out the complete read-across exercise for at least one chemical category per read-across scenario in the context of SEURAT-1, and the results of this exercise will be completed and presented by the end of the research initiative in December 2015.

Computer models versus reality: how well do in silico models currently predict the sensitization potential of a substance.

National legislations for the assessment of the skin sensitization potential of chemicals are increasingly based on the globally harmonized system (GHS). In this study, experimental data on 55 non-sensitizing and 45 sensitizing chemicals were evaluated according to GHS criteria and used to test the performance of computer (in silico) models for the prediction of skin sensitization. Statistic models (Vega, Case Ultra, TOPKAT), mechanistic models (Toxtree, OECD (Q)SAR toolbox, DEREK) or a hybrid model (TIMES-SS) were evaluated. Between three and nine of the substances evaluated were found in the individual training sets of various models. Mechanism based models performed better than statistical models and gave better predictivities depending on the stringency of the domain definition. Best performance was achieved by TIMES-SS, with a perfect prediction, whereby only 16% of the substances were within its reliability domain. Some models offer modules for potency; however predictions did not correlate well with the GHS sensitization subcategory derived from the experimental data. In conclusion, although mechanistic models can be used to a certain degree under well-defined conditions, at the present, the in silico models are not sufficiently accurate for broad application to predict skin sensitization potentials.

Structural basis and targeting of the interaction between fibroblast growth factor-inducible 14 and tumor necrosis factor-like weak inducer of apoptosis.

Deregulation of the TNF-like weak inducer of apoptosis (TWEAK)-fibroblast growth factor-inducible 14 (Fn14) signaling pathway is observed in many diseases, including inflammation, autoimmune diseases, and cancer. Activation of Fn14 signaling by TWEAK binding triggers cell invasion and survival and therefore represents an attractive pathway for therapeutic intervention. Based on structural studies of the TWEAK-binding cysteine-rich domain of Fn14, several homology models of TWEAK were built to investigate plausible modes of TWEAK-Fn14 interaction. Two promising models, centered on different anchoring residues of TWEAK (tyrosine 176 and tryptophan 231), were prioritized using a data-driven strategy. Site-directed mutagenesis of TWEAK at Tyr(176), but not Trp(231), resulted in the loss of TWEAK binding to Fn14 substantiating Tyr(176) as the anchoring residue. Importantly, mutation of TWEAK at Tyr(176) did not disrupt TWEAK trimerization but failed to induce Fn14-mediated nuclear factor κ-light chain enhancer of activated B cell (NF-κB) signaling. The validated structural models were utilized in a virtual screen to design a targeted library of small molecules predicted to disrupt the TWEAK-Fn14 interaction. 129 small molecules were screened iteratively, with identification of molecules producing up to 37% inhibition of TWEAK-Fn14 binding. In summary, we present a data-driven in silico study revealing key structural elements of the TWEAK-Fn14 interaction, followed by experimental validation, serving as a guide for the design of small molecule inhibitors of the TWEAK-Fn14 ligand-receptor interaction. Our results validate the TWEAK-Fn14 interaction as a chemically tractable target and provide the foundation for further exploration utilizing chemical biology approaches focusing on validating this system as a therapeutic target in invasive cancers.

Rough Set Theory as an Interpretable Method for Predicting the Inhibition of Cytochrome P450 1A2 and 2D6.

Early prediction of ADME properties such as the cytochrome P450 (CYP) mediated drug-drug interactions is an important challenge in the drug discovery area. In this study, we propose to couple an original data mining approach based on Rough Set Theory (RST) to a structural description of molecules. The latter was achieved by using two types of structural keys: (1) the MACCS keys and (2) a set of five in-house fingerprints based on properties of the electron density distributions of chemical groups. The compounds considered are involved in the inhibition of CYP1A2 and CYP2D6. RST allowed the extraction of rules further used as classifiers to predict the inhibitory profile of an independent set of molecules. The results reached prediction accuracies of 90.6 and 88.2 % for CYP1A2 and CYP2D6, respectively. In addition, these classifiers were analyzed to determine which structural fragments were most used for building the rules, revealing relationships between the occurrence of particular molecular fragments and CYP inhibition. The results assessed RST as a suitable tool to build strongly predictive models and infer structure-activity rules associated with potency.

Virtual screening for cytochromes p450: successes of machine learning filters.

Cytochromes P450 (CYPs) are crucial targets when predicting the ADME properties (absorption, distribution, metabolism, and excretion) of drugs in development. Particularly, CYPs mediated drug-drug interactions are responsible for major failures in the drug design process. Accurate and robust screening filters are thus needed to predict interactions of potent compounds with CYPs as early as possible in the process. In recent years, more and more 3D structures of various CYP isoforms have been solved, opening the gate of accurate structure-based studies of interactions. Nevertheless, the ligand-based approach still remains popular. This success can be explained by the growing number of available data and the satisfying performances of existing machine learning (ML) methods. The aim of this contribution is to give an overview of the recent achievements in ML applications to CYP datasets. Particularly, popular methods such as support vector machine, decision trees, artificial neural networks, k-nearest neighbors, and partial least squares will be compared as well as the quality of the datasets and the descriptors used. Consensus of different methods will also be discussed. Often reaching 90% of accuracy, the models will be analyzed to highlight the key descriptors permitting the good prediction of CYPs binding.

Can descriptors of the electron density distribution help to distinguish functional groups?

Our study is aimed at understanding the characteristics of functional group descriptors based on peaks of the electronic density distribution rho(->r) . The descriptors calculated are the rho(->r) value at peak location, volume, ellipticity, curvatures of rho ( r) , and the peak-functional group distance. By the implementation of an automated and global process for large-scale calculation of the descriptors, we generated a statistically meaningful data set focusing on the association between peaks and 77 types of functional groups extracted from 62,936 organic molecules issued from the Cambridge Structural Database. Statistical analyses demonstrated that selected descriptors are capable of discriminating subtypes of functional groups. A projection in a principal component space coupled to a hierarchical clustering confirmed the suitability of the descriptors to provide an appropriate description of the functional groups. The results indicated that functional similarity or dissimilarity could be quantified based on electron density descriptors.

Recursive partitioning for the prediction of cytochromes P450 2D6 and 1A2 inhibition: importance of the quality of the dataset.

The purpose of this study was to explore the use of detailed biological data in combination with a statistical learning method for predicting the CYP1A2 and CYP2D6 inhibition. Data were extracted from the Aureus-Pharma highly structured databases which contain precise measures and detailed experimental protocol concerning the inhibition of the two cytochromes. The methodology used was Recursive Partitioning, an easy and quick method to implement. The building of models was preceded by the evaluation of the chemical space covered by the datasets. The descriptors used are available in the MOE software suite. The models reached at least 80% of Accuracy and often exceeded this percentage for the Sensitivity (Recall), Specificity, and Precision parameters. CYP2D6 datasets provided 11 models with Accuracy over 80%, while CYP1A2 datasets counted 5 high-accuracy models. Our models can be useful to predict the ADME properties during the drug discovery process and are indicated for high-throughput screening.