Use Of Quantitative Structure-activity Relationships Research Articles

Integrating Genetic Algorithm and LightGBM for QSAR Modeling of Acetylcholinesterase Inhibitors in Alzheimer's Disease Drug Discovery

This study explores the use of Quantitative Structure-Activity Relationship (QSAR) studies using genetic algorithm (GA) and LightGBM to search for acetylcholinesterase (AChE) inhibitors for Alzheimer's disease. The study uses a dataset of 6,157 AChE inhibitors and their IC50 values. A LightGBM model is trained and evaluated for classification performance. The results show that the LightGBM model achieved high performance on the training and testing set, with an accuracy of 92.49% and 82.47%, respectively. This study demonstrates the potential of GA and LightGBM in the drug discovery process for AChE inhibitors in Alzheimer's disease. The findings contribute to the drug discovery process by providing insights about AChE inhibitors that allow more efficient screening of potential compounds and accelerate the identification of promising candidates for development and therapeutic use.

Multi-Strategy Assessment of Different Uses of QSAR under REACH Analysis of Alternatives to Advance Information Transparency.

Under the Registration, Evaluation, Authorization, and Restriction of Chemicals (REACH) analysis of alternatives (AoA) process, quantitative structure–activity relationship (QSAR) models play an important role in expanding information gathering and organizing frameworks. Increasingly recognized as an alternative to testing under registration. QSARs have become a relevant tool in bridging data gaps and supporting weight of evidence (WoE) when assessing alternative substances. Additionally, QSARs are growing in importance in integrated testing strategies (ITS). For example, the REACH ITS framework for specific endpoints directs registrants to consider non-testing results, including QSAR predictions, when deciding if further animal testing is needed. Despite the raised profile of QSARs in these frameworks, a gap exists in the evaluation of QSAR use and QSAR documentation under authorization. An assessment of the different uses (e.g., WoE and ITS) in which QSAR predictions play a role in evidence gathering and organizing remains unaddressed for AoA. This study approached the disparity in information for QSAR predictions by conducting a substantive review of 24 AoA through May 2017, which contained higher-tier endpoints under REACH. Understanding the manner in which applicants manage QSAR prediction information in AoA and assessing their potential within ITS will be valuable in promoting regulatory use of QSARs and building out future platforms in the face of rapidly evolving technology while advancing information transparency.

Outliers in SAR and QSAR: 4. effects of allosteric protein-ligand interactions on the classical quantitative structure-activity relationships.

Effects of allosteric interactions on the classical structure-activity relationship (SAR) and quantitative SAR (QSAR) have been investigated. Apprehending the outliers in SAR and QSAR studies can improve the quality, predictability, and use of QSAR in designing unknown compounds in drug discovery research. We explored allosteric protein-ligand interactions as a possible source of outliers in SAR/QSAR. We used glycogen phosphorylase as an example of a protein that has an allosteric site. Examination of the ligand-bound x-ray crystal structures of glycogen phosphorylase revealed that many inhibitors bound at more than one binding site. The results of QSAR analyses of the inhibitors included a QSAR that recognized an outlier bound at a distinctive allosteric binding site. The case provided an example of constructive use of QSAR identifying outliers with alternative binding modes. Other allosteric QSARs that captured our attention were the inverted parabola/bilinear QSARs. The x-ray crystal structures and the QSAR analyses indicated that the inverted parabola QSARs could be associated with the conformational changes in the allosteric interactions. Our results showed that the normal parabola, as well as the inverted parabola QSARs, can describe the allosteric interactions. Examination of the ligand-bound X-ray crystal structures of glycogen phosphorylase revealed that many inhibitors bound at more than one binding site. The results of QSAR analyses of the inhibitors included a QSAR that recognized an outlier bound at a distinctive allosteric binding site.

A Combined approach of QSAR study, molecular docking and pharmacokinetics prediction of promising Amide-Ac6-aminoacetonitriles Cathepsin K inhibitors

• Combined use of QSAR, molecular docking and pharmacokinetics prediction approaches. • One promising Cathepsin K inhibitor was identified. Cathepsin K (Cat-K) highly expressed in osteoclasts, is considered to be a promising target for osteoclast-related maladies. In the present study, quantitative structure–activity relationships (QSAR) and molecular docking approaches were used to identify promising Cat-K inhibitors from a series of Amide-Ac6-aminoacetonitriles analogs. During the two-dimensional QSAR study, the statistical model PLS_MDL was chosen (r 2 = 0.823, q 2 (cross-validation) = 0.732). Descriptors with positive or negative contributions were derived from model PLS_MDL. In order to build robust three-dimensional QSAR models, three different molecular fingerprints were used as analysis precepts for clustering and three different minimization algorithms were employed for generation of molecular poses. The best database which used molecular fingerprint of FCFP and minimization algorithm of Smart Minimizer was selected for further QSAR model construction. The ideal CoMFA and CoMSIA models were acquired (CoMFA: r 2 = 0.963, q 2 = 0.674, r 2 pred = 0.702 and Q 2 (F2) = 0.616; CoMSIA_SEHA: r 2 = 0.989, q 2 = 0.622, r 2 pred = 0.719 and Q 2 (F2) = 0.638). Physicochemical, structural, electrostatic, and steric properties were derived from the created QSAR models and then applied to portray a design scheme. Furthermore, after a selecting of docking scheme, docking studies were carried out to measure the binding affinity of eligible query compounds and to reveal the detailed interactions between ligand and the target protein. Amongst these designed compounds, compound19OX32 was focused for its preferable predicted activity and docking score. Strong interactions and extra hydrogen bonds which echoed the previous summarized design scheme were also observed via a deep investigation of docking result. Consequently, compound 19OX32 was identified as promising Cat-K inhibitor.

Quantitative Structure-Activity Relationship (QSAR) Studies on the Toxic Effects of Nitroaromatic Compounds (NACs): A Systematic Review.

Nitroaromatic compounds (NACs) are ubiquitous in the environment due to their extensive industrial applications. The recalcitrance of NACs causes their arduous degradation, subsequently bringing about potential threats to human health and environmental safety. The problem of how to effectively predict the toxicity of NACs has drawn public concern over time. Quantitative structure–activity relationship (QSAR) is introduced as a cost-effective tool to quantitatively predict the toxicity of toxicants. Both OECD (Organization for Economic Co-operation and Development) and REACH (Registration, Evaluation and Authorization of Chemicals) legislation have promoted the use of QSAR as it can significantly reduce living animal testing. Although numerous QSAR studies have been conducted to evaluate the toxicity of NACs, systematic reviews related to the QSAR modeling of NACs toxicity are less reported. The purpose of this review is to provide a thorough summary of recent QSAR studies on the toxic effects of NACs according to the corresponding classes of toxic response endpoints.

Application of QSAR Method in the Design of Enhanced Antimalarial Derivatives of Azetidine-2-carbonitriles, their Molecular Docking, Drug-likeness, and SwissADME Properties.

The resistance of the P. falciparum strain to some of the antimalarial drugs has been a dominant dilemma facing the treatment of this fetid disease. This necessitates the detection and development of new antimalarial agents targeting the P. falciparum. Azetidine-2-carbonitriles reported for its antimalarial activities, could provide an alternative to the customized antimalarial drugs. Leading to the use of quantitative structure-activity relationship (QSAR) studies, which relates the structures of Azetidine-2-carbonitriles with their activities to generate predictive models. The structures were optimized using density functional theory (DFT) DFT/B3LYP/6-31G* basis set to generate their molecular descriptors, where five predictive models were constructed using the generated descriptors. The models were constructed using the genetic function algorithm component of a material studio, where the model with good statistical parameters, high coefficient of determination (R2) = 0.9465, cross-validated R2 (Q2cv) = 0.8981, Q2 (L4O)cv = 0.9272, and highest external validated R2 (R2pred) = 0.6915 was selected as the best model. These statistical results show the robustness, excellent power of prediction, and validity of the selected model. The descriptor, SpMax2_Bhp (the maximum absolute eigenvalue of Barysz matrix for n = 2 was weighted by polarizability), was revealed to be the most influential in the model due to its highest mean effect. The descriptor played a role in the design of sixteen (16) theoretical derivatives of Azetidine-2-carbonitriles using compound 25 as the design template by increasing polarizability of the compounds through substitution of the various group with electron deactivating groups (F, I, Cl, SO3H, CN, NO2, etc.) at different position of the template. The designed compounds were docked with Plasmodium falciparum dihydroorotate dehydrogenase (Pf-DHODH), giving compound D9 the highest binding energy. The designed compounds were further screened for their drug-likeness, where they all pass Lipinski’s RO5. All the compounds show good skin permeability coefficient and have low Gastrointestinal absorption while few compounds D1, D2, D3, D14, and D15 inhibiting the CYP1A2.

QSAR Analysis of Selected Antimicrobial Structures Belonging to Nitro-derivatives of Heterocyclic Compounds

Background: Nitro-derivatives of heterocyclic compounds were used as active agents against pathogenic microorganisms. A set of 4- and 5-nitroimidazole derivatives exhibiting antimicrobial activity was analyzed with the use of Quantitative Structure-Activity Relationships (QSAR) method. The study included compounds used both in documented treatment and those described as experimental. Objective: The purpose of this study was to demonstrate the common and differentiating characteristics of the above-mentioned chemical compounds alike physicochemically as well as pharmacologically based on the quantum chemical calculations and microbiological activity data. Methods: During the study PCA and MLR analysis were performed, as the types of proposed chemometric approach. The semi-empirical and ab initio level of in silico molecular modeling was performed for calculations of molecular descriptors. Results: QSAR models were proposed based on chosen descriptors. The relationship between the nitro-derivatives structure and microbiological activity data was able to class and describe the antimicrobial activity with the use of statistically significant molecular descriptors. Conclusion: The applied chemometric approaches revealed the influential features of the tested structures responsible for the antimicrobial activity of studied nitro-derivatives.

Activity Evaluation and Selection of Some Classes of Antibiotics with the use of Semi-Empirical Quantum Mechanics and Quantitative Structure- Activity Relationships Approach.

A set of β-lactam antibiotics, aminoglycoside antibiotics, and tetracycline antibiotics were proposed and analyzed with the use of Quantitative Structure-Activity Relationships (QSAR) method. The characterization of selected antimicrobial compounds in terms of both physicochemical and pharmacological on the basis of calculations of quantum mechanics and possessed biological activity data. During the study, Multiple Linear Regression (MLR) supported with Factor Analysis (FA) and Principal Component Analysis (PCA) was made, as the types of proposed chemometric approach; the semi-empirical level of in silico molecular modeling was used for calculations and comparison of molecular descriptors both in a vacuum and in the aquatic environment. The relationships between structure and microbiological activity enabled the characterization and description of the analyzed molecules using statistically significant descriptors belonging in most cases to different structural, geometric and electronic elements defining at the same time the properties of the studied three different classes of examined antibiotics. The chemometric methods used revealed the influence of some of the elements of structures examined molecules belonging to main antibiotics classes and responsible for the antimicrobial activity.

Methods of Determining Retardation Coefficients of Organic Compounds in Aquifers

Sorption is an important mechanism that affects the mobility of organic compounds in the subsurface environment. Sorbed compounds move slower than that of groundwater, causing retardation in their movement. Accurate determination of retardation coefficients (R) of organic compounds in aquifers is critical for understanding their movement, fate, and remediation. Several methods, including predictive tools, laboratory experiments, and field experiments have been utilized for determining sorption-related retardation. The objective of this paper was to review and compare between the different methods used for the determination of R of organic compounds, with emphasis on predictive- and laboratory-based approaches. Predictive tools are based on the use of quantitative structure-activity relationships (QSARs). Laboratory methods utilize different types of reactors including batch, stirred-flow, circulation-through-column, or miscible displacement through packed columns. In addition, data from the column method have been analyzed in various ways to determine R. Discrepancies between results from different methods or from different analysis approaches have been reported. This create uncertainty about the suitability of some of these methods or the used analysis approaches. This paper highlights the possible causes for the observed discrepancy and establishes the limitations and appropriateness of the used methods and analysis approaches.

Improvement of quantitative structure-activity relationship (QSAR) tools for predicting Ames mutagenicity: outcomes of the Ames/QSAR International Challenge Project.

The International Conference on Harmonization (ICH) M7 guideline allows the use of in silico approaches for predicting Ames mutagenicity for the initial assessment of impurities in pharmaceuticals. This is the first international guideline that addresses the use of quantitative structure–activity relationship (QSAR) models in lieu of actual toxicological studies for human health assessment. Therefore, QSAR models for Ames mutagenicity now require higher predictive power for identifying mutagenic chemicals. To increase the predictive power of QSAR models, larger experimental datasets from reliable sources are required. The Division of Genetics and Mutagenesis, National Institute of Health Sciences (DGM/NIHS) of Japan recently established a unique proprietary Ames mutagenicity database containing 12140 new chemicals that have not been previously used for developing QSAR models. The DGM/NIHS provided this Ames database to QSAR vendors to validate and improve their QSAR tools. The Ames/QSAR International Challenge Project was initiated in 2014 with 12 QSAR vendors testing 17 QSAR tools against these compounds in three phases. We now present the final results. All tools were considerably improved by participation in this project. Most tools achieved >50% sensitivity (positive prediction among all Ames positives) and predictive power (accuracy) was as high as 80%, almost equivalent to the inter-laboratory reproducibility of Ames tests. To further increase the predictive power of QSAR tools, accumulation of additional Ames test data is required as well as re-evaluation of some previous Ames test results. Indeed, some Ames-positive or Ames-negative chemicals may have previously been incorrectly classified because of methodological weakness, resulting in false-positive or false-negative predictions by QSAR tools. These incorrect data hamper prediction and are a source of noise in the development of QSAR models. It is thus essential to establish a large benchmark database consisting only of well-validated Ames test results to build more accurate QSAR models.

Data-based review of QSARs for predicting genotoxicity: the state of the art.

With the aim of providing faster, more economical, animal-free tools to predict toxicity, quantitative structure-activity relationships (QSAR) approaches are increasingly applied to the chemical risk assessment-in particular genotoxicity and carcinogenicity. The more recent period of time has witnessed refinements of the predictive systems, with the collection of larger training sets and continued fine-tuning, together with an increased interest for the use of QSAR by regulatory authorities. This literature review provides an updated snapshot of the present state of the art in the evaluation of QSAR methods as applied to genotoxicity. Overall, the abilities of software tools to predict Ames test mutagenicity were comparable with previously published evaluations, with sensitivity ranging 0.72-0.96, and specificity ranging 0.65-0.86 in applications to public data sets. These values compare quite fairly with the intrinsic variability of the Ames test. A preliminary analysis indicated a consistency with the results of the Japan Division of Genetics and Mutagenesis, National Institute of Health Sciences of Japan (DGM/NIHS) Ames/QSAR international collaborative project. Applications to a variety of external test sets pointed to the need of further improvements of the coverage/representation of the whole chemical space. Combinations of tools showed that sensitivity is usually increased at the expense of a decrease in specificity, whereas the supervision by expert judgement generated more equilibrated results. This points to the existence of a large area of context-dependent expert knowledge, which has not been formalised yet and has the potential to substantially improve the prediction systems. The overall evidence suggests that (Q)SARs for the Ames test have sufficient reliability for use in prioritisation processes as well as to support regulatory decisions in combination with other evidence. The use of highly sensitive genotoxicity QSARs in tiered integrations with other tools is suggested as a mean to shortlist chemicals for which no further testing is necessary.

The potential to use QSAR to populate ecotoxicity characterisation factors for simplified LCIA and chemical prioritisation

PurposeToday’s chemical society use and emit an enormous number of different, potentially ecotoxic, chemicals to the environment. The vast majority of substances do not have characterisation factors describing their ecotoxicity potential. A first stage, high throughput, screening tool is needed for prioritisation of which substances need further measures.MethodsUSEtox characterisation factors were calculated in this work based on data generated by quantitative structure-activity relationship (QSAR) models to expand substance coverage where characterisation factors were missing. Existing QSAR models for physico-chemical data and ecotoxicity were used, and to further fill data gaps, an algae QSAR model was developed. The existing USEtox characterisation factors were used as reference to evaluate the impact from the use of QSARs to generate input data to USEtox, with focus on ecotoxicity data. An inventory of chemicals that make up the Swedish societal stock of plastic additives, and their associated predicted emissions, was used as a case study to rank chemicals according to their ecotoxicity potential.Results and discussionFor the 210 chemicals in the inventory, only 41 had characterisation factors in the USEtox database. With the use of QSAR generated substance data, an additional 89 characterisation factors could be calculated, substantially improving substance coverage in the ranking. The choice of QSAR model was shown to be important for the reliability of the results, but also with the best correlated model results, the discrepancies between characterisation factors based on estimated data and experimental data were very large.ConclusionsThe use of QSAR estimated data as basis for calculation of characterisation factors, and the further use of those factors for ranking based on ecotoxicity potential, was assessed as a feasible way to gather substance data for large datasets. However, further research and development of the guidance on how to make use of estimated data is needed to achieve improvement of the accuracy of the results.

The OECD Principles for (Q)SAR Models in the Context of Knowledge Discovery in Databases (KDD).

The steps followed in the knowledge discovery in databases (KDD) process are well documented and are widely used in different areas where exploration of data is used for decision making. In turn, while different workflows for developing quantitative structure-activity relationship (QSAR) models have been proposed, including combinatorial use of QSAR, there is now agreement on common requirements for building trustable predictive models. In this work, we analyze and confront the steps involved in KDD and QSAR and present how they comply with the OECD principles for the validation, for regulatory purposes, of QSAR models.

Regulation of engineered nanomaterials: current challenges, insights and future directions.

Substantial production and wide applications of engineered nanomaterials (ENMs) have raised concerns over their potential influences on the environment and humans. However, regulations of products containing ENMs are scarce, even in countries with the greatest volume of ENMs produced, such as the United States and China. After a comprehensive review of life cycles of ENMs, five major challenges to regulators posed by ENMs are proposed in this review: (a) ENMs exhibit variable physicochemical characteristics, which makes them difficult for regulators to establish regulatory definition; (b) Due to diverse sources and transport pathways for ENMs, it is difficult to monitor or predict their fates in the environment; (c) There is a lack of reliable techniques for quantifying exposures to ENMs; (d) Because of diverse intrinsic properties of ENMs and dynamic environmental conditions, it is difficult to predict bioavailability of ENMs on wildlife and the environment; and (e) There are knowledge gaps in toxicity and toxic mechanisms of ENMs from which to predict their hazards. These challenges are all related to issues in conventional assessments of risks that regulators rely on. To address the fast-growing nanotechnology market with limited resources, four ENMs (nanoparticles of Ag, TiO2, ZnO and Fe2O3) have been prioritized for research. Compulsory reporting schemes (registration and labelling) for commercial products containing ENMs should be adopted. Moreover, to accommodate their potential risks in time, an integrative use of quantitative structure-activity relationship and adverse outcome pathway (QSAR-AOP), together with qualitative alternatives to conventional risk assessment are proposed as tools for decision making of regulators.

Towards Predicting the Cytochrome P450 Modulation: From QSAR to Proteochemometric Modeling.

Drug metabolism determines the fate of a drug when it enters the human body and is a critical factor in defining their absorption, distribution, metabolism, excretion and toxicity (ADMET) characteristics. Among the various drug metabolizing enzymes, cytochrome P450s (CYP450) constitute an important protein family that aside from functioning in xenobiotic metabolism, is also responsible for a diverse array of other roles encompassing steroid and cholesterol biosynthesis, fatty acid metabolism, calcium homeostasis, neuroendocrine functions and growth regulation. Although CYP450 typically converts xenobiotics into safe metabolites, there are some situations whereby the metabolite is more toxic than its parent molecule. Computational modeling has been instrumental in CYP450 research by rationalizing the nature of the binding event (i.e. inhibit or induce CYP450s) or metabolic stability of query compounds of interest. A plethora of computational approaches encompassing ligand, structure and systems based approaches have been utilized to model CYP450-ligand interactions. This review provides a brief background on the CYP450 family (i.e. its roles, advantages and disadvantages as well as its modulators) and then discusses the various computational approaches that have been used to model CYP450-ligand interaction. Particular focus was given to the use of quantitative structure-activity relationship (QSAR) and more recent proteochemometric modeling studies. Finally, a perspective on the current state of the art and future trends of the field is also provided.

Acute aquatic toxicity of organic solvents modeled by QSARs.

To limit in vivo experiments, the use of quantitative structure-activity relationships (QSARs) is advocated by REACH regulation to predict the required fish, invertebrate, and algae EC50 for chemical registration. The aim of this work was to develop reliable QSARs in order to model both invertebrate and algae EC50 for organic solvents, regardless of the mechanism of toxic action involved. EC50 represents the concentration producing the 50% immobilization of invertebrates or the 50% growth inhibition of algae. The dataset was composed of 122 organic solvents chemically heterogeneous which were characterized by their invertebrate and/or algae EC50. These solvents were described by physico-chemical descriptors and quantum theoretical parameters calculated via density functional theory. QSAR models were developed by multiple linear regression using the ordinary least squares method and descriptor selection was performed by the Kubinyi function. Invertebrate EC50 was well-described with LogP, dielectric constant, surface tension, and minimal atomic Mulliken charges while algae EC50 of organic solvents (except amines) was predicted with LogP and LUMO energy. To evaluate robustness and predictive performance of the QSARs developed, several strategies have been used to select solvent training sets (random, EC50-based selection and a space-filling design) and both internal and external validations were performed.

The utility of QSARs in predicting acute fish toxicity of pesticide metabolites: A retrospective validation approach

The European Plant Protection Products Regulation 1107/2009 requires that registrants establish whether pesticide metabolites pose a risk to the environment. Fish acute toxicity assessments may be carried out to this end. Considering the total number of pesticide (re-) registrations, the number of metabolites can be considerable, and therefore this testing could use many vertebrates. EFSA's recent “Guidance on tiered risk assessment for plant protection products for aquatic organisms in edge-of-field surface waters” outlines opportunities to apply non-testing methods, such as Quantitative Structure Activity Relationship (QSAR) models. However, a scientific evidence base is necessary to support the use of QSARs in predicting acute fish toxicity of pesticide metabolites. Widespread application and subsequent regulatory acceptance of such an approach would reduce the numbers of animals used.The work presented here intends to provide this evidence base, by means of retrospective data analysis. Experimental fish LC50 values for 150 metabolites were extracted from the Pesticide Properties Database (http://sitem.herts.ac.uk/aeru/ppdb/en/atoz.htm). QSAR calculations were performed to predict fish acute toxicity values for these metabolites using the US EPA's ECOSAR software. The most conservative predicted LC50 values generated by ECOSAR were compared with experimental LC50 values.There was a significant correlation between predicted and experimental fish LC50 values (Spearman rs = 0.6304, p < 0.0001). For 62% of metabolites assessed, the QSAR predicted values are equal to or lower than their respective experimental values. Refined analysis, taking into account data quality and experimental variation considerations increases the proportion of sufficiently predictive estimates to 91%. For eight of the nine outliers, there are plausible explanation(s) for the disparity between measured and predicted LC50 values.Following detailed consideration of the robustness of this non-testing approach, it can be concluded there is a strong data driven rationale for the applicability of QSAR models in the metabolite assessment scheme recommended by EFSA. As such there is value in further refining this approach, to improve the method and enable its future incorporation into regulatory guidance and practice.

The utility of QSARs in predicting acute fish toxicity of pesticide metabolites: A retrospective validation approach

The European Plant Protection Products Regulation 1107/2009 requires that registrants establish whether pesticide metabolites pose a risk to the environment. Fish acute toxicity assessments may be carried out to this end. Considering the total number of pesticide (re-) registrations, the number of metabolites can be considerable, and therefore this testing could use many vertebrates. EFSA's recent Guidance on tiered risk assessment for plant protection products for aquatic organisms in edge-of-field surface waters outlines opportunities to apply non-testing methods, such as Quantitative Structure Activity Relationship (QSAR) models. However, a scientific evidence base is necessary to support the use of QSARs in predicting acute fish toxicity of pesticide metabolites. Widespread application and subsequent regulatory acceptance of such an approach would reduce the numbers of animals used. The work presented here intends to provide this evidence base, by means of retrospective data analysis. Experimental fish LC50 values for 150 metabolites were extracted from the Pesticide Properties Database (http://sitem.herts.ac.uk/aeru/ppdb/en/atoz.htm). QSAR calculations were performed to predict fish acute toxicity values for these metabolites using the US EPA's ECOSAR software. The most conservative predicted LC50 values generated by ECOSAR were compared with experimental LC50 values. There was a significant correlation between predicted and experimental fish LC50 values (Spearman rs = 0.6304, p < 0.0001). For 62% of metabolites assessed, the QSAR predicted values are equal to or lower than their respective experimental values. Refined analysis, taking into account data quality and experimental variation considerations increases the proportion of sufficiently predictive estimates to 91%. For eight of the nine outliers, there are plausible explanation(s) for the disparity between measured and predicted LC50 values. Following detailed consideration of the robustness of this non-testing approach, it can be concluded there is a strong data driven rationale for the applicability of QSAR models in the metabolite assessment scheme recommended by EFSA. As such there is value in further refining this approach, to improve the method and enable its future incorporation into regulatory guidance and practice.

Prediction of immobilised artificial membrane chromatography retention factors using theoretical molecular fragments and structural features

Many in silico alternatives to aquatic toxicity tests rely on hydrophobicity-based quantitative structure–activity relationships (QSARs). Hydrophobicity is often estimated as log P, where P is the octanol–water partition coefficient. Immobilised artificial membrane (IAM) high performance liquid chromatography (HPLC) may be a more biologically relevant alternative to log P. The aim of this study was to investigate the applicability of a theoretical structural fragment and feature-based method to predict log k IAM (the logarithm of the retention index determined by IAM–HPLC) values. This will allow the prediction of log k IAM based on chemical structure alone. The use of structural fragment values to predict log P was first proposed in the 1970s. The application of a similar method using fragment values to predict log k IAM is a novel approach. Values of log k IAM were determined for 22 aliphatic and 42 aromatic compounds using an optimised and robust IAM–HPLC assay. The method developed shows good predictive performance using leave-one-out cross validation and application to an external validation set not seen a priori by the training set also generated good predictive values. The ability to predict log k IAM without the need for practical measurement will allow for the increased use of QSARs based on this descriptor.

Graph Machine Based-QSAR Approach for Modeling Thermodynamic Properties of Amines: Application to CO2Capture in Postcombustion

Amine scrubbing is usually considered as the most efficient technology for CO<sub>2<sub/> mitigation through postcombustion Carbon Capture and Storage (CCS). However, optimization of the amine structure to improve the solvent properties requires to sample a large number of possible candidates and hence to gather a large amount of experimental data. In this context, the use of QSAR (Quantitative Structure Activity Relationship) statistical modeling is a powerful tool as it performs a mapping of a set of input vectors (i.e. the characteristics or the properties of the molecules under consideration) to a set of output vectors (i.e. their targeted properties). In this work, we used a high throughput screening experimental device to measure CO<sub>2<sub/> solubility data on a set of 46 amine aqueous solutions. Absorption isotherms are represented using a thermodynamic model based on two thermodynamic constants, pKa<sup>*<sup/> and pKc<sup>*<sup/> , accounting for the main chemical reactions occurring in the liquid phase between amine and CO<sub>2<sub/>. Then, we used a statistical approach named Graph Machines at the same time to cluster the molecules and to model the variation of the acidity constant pKa<sup>*<sup/> as a function of the molecular structure. The originality of our approach is the use of graphs to represent molecules in multidimensional spaces and simultaneously construct predictive models of their physicochemical properties based on these graphs. This approach is applied in this paper to predict the thermodynamic properties of a set of 5 new molecules.