The concern to protect human health and the environment has prompted significant changes in EU regulation on chemical substances. The European Chemicals Agency (ECHA) plays the role in implementing the Registration, Evaluation, Authorization and Restriction of Chemicals (REACH) legislation, that requires industry to evaluate the toxicity of chemical substances that are in use but have never been subjected to regulatory testing. REACH regulation has also raised strong criticism and concern from society and industrials because of ethical and economic reasons. The toxicity evaluation of chemicals requires costly, time-consuming and ethically questionable animal experiments. As consequence, this European regulation promotes scientific innovation and encourages the use of results generated by alternative methods, including especially non-testing methods (NTMs), also referred to as in silico tools. Among them, “Quantitative Structure-Activity Relationships” (QSAR) methods are one of the most recognized machine learning methods in drug design, toxicology, industrial and environmental chemistry. Nowadays, they can be included in integrated testing strategies (ITS), to provide information for hazard and risk assessment, classification and labelling. We propose the development of an ensemble of QSAR predictive models for several parameters related with the different kinds of genotoxicity damage. These chemoinformatic will be implemented on a proprietary computational technological platform. Particular attention will be also payed to the generation of models for nanomaterials, taking into account their high and growing impact nowadays on industry in general.The QSAR models and integration algorithms will be characterized by their reliability, and will be developed according to the rules set out by the OECD, therefore guaranteeing their validity in REACH.

Nanotechnology is one of the fastest growing and most promising technologies in our society (Forster et al. 2011), promoting the development a new generation of smart and innovative products and processes that have created tremendous growth potential for a large number of industry sectors such as composites, colouring, ceramics, electronics, nutrition, cosmetics, energy, optics, automotive, as well as numerous other industrial sectors. Currently, there is a need of ensuring a safe and sustainable development of the nanotechnology, which implies a better understanding of the potential harmful effects that ENMs may have on human´s health or the environment. New paradigms are necessary to identify high concern ENMs and predict relevant endpoints for risk assessment, reducing the cost and timescale derived from the use of in vivo or in vitro assays. QSAR approaches have only recently been used to predict biological effects of ENMs, with only few Quantitative Nano- Structure Activity Relationships models described in the literature. The lack of available data explains why there is almost no literature reporting the use of computational modelling techniques applied to ENMs, especially in the area of nanotoxicology. On the other hand, current toxicological regulation, such as the Registration, Evaluation, Authorisation and Restriction of Chemicals (REACH), strongly promotes the use of these predictive modelling. On the basis of the concept of the project, the main objective of the Nano-QSAR project is to develop new scientifically validated QSARs models to predict REACH relevant toxicological, ecotoxicological and environmental endpoints of a priority list of ENMs such as Metal Oxide Nanoparticles (MOx) and Quantum Dots (QD) on the basis of available literature and own experimental data.

The main goal of the proposed research project is the computational evaluation of eco-toxicity (diverse endpoints) of various chemicals that are vastly utilized and produced by the pharmaceutical and cosmetic industries, such as green solvents (including future ones, i.e., ionic liquids and deep eutectic solvents) and active pharmaceutical ingredients (API). We will be majorly focusing on toxicity in aquatic environment, where the toxicity data will cover four trophic levels of aquatic organisms, i.e., fish (vertebrates), invertebrates such as daphnids, algae (aquatic plants), and microorganisms. The toxicity related properties that will be studied include acute and chronic toxicity, biodegradation and bioaccumulation. The research methodology to perform toxicity assessment and for understanding the structural features responsible for the eco-toxicity, will involve diverse Artificial Intelligence (AI) and chemoinformatics techniques like Quantitative Structure-Activity Relationship (QSAR), interspecies QSAR (QAAR), toxicophore mapping, virtual screening, similarity search, clustering techniques, multimedia mass-balance (MM) modeling (to understand the distribution profile of chemicals in different environmental compartments), matched molecular pair (MMPs) analysis etc. The knowledge gained from the study will help in classifying existing chemicals into toxic and non-toxic groups and will also help in designing novel analogues of selected chemical that will show better desirable physicochemical properties with less or no eco-toxicity. This project will also include development of AI software tools and scheming KNIME workflows for various computational tasks.

Chemicals substances are presented in daily life like cosmetics, functional food or agro-industry. The EC Regulation 1907/2006 for Registration, Evaluation, Authorization and Restriction of Chemicals (REACH), for the production and use of chemicals when they are produced/imported in/to Europe in an amount greater than 1 ton/year has two major and clear customer paint points: first, REACH implies very high costs due to the experimental and administrative work required, which represents a particularly sensitive issue for SMEs. Second, REACH implies the need for a huge number of animal tests to demonstrate the safety of chemicals under approval. Animal experiments require much time for preparation and execution, and are expensive and ethically questionable. Instead, QSAR computational models represent significant savings in time, resources and money, since the applicability of predictive models is ease and immediate. And this business need from the European chemical industry SMEs is our market opportunity, by the final development of a novel high scalable technological platform of easy access to experienced and non-experienced end-users. These platform will include QSAR models of (eco)toxicology of chemicals -including nanomaterials (NMs)- to comply with the strict EC Regulation, where assays with animals in some specific domains, such as cosmetic, are completely forbidden. This customer self-using platform will include an additional tool for the evaluation of the appropriateness of replacing the chemicals studied for alternative compounds which a safer (eco)toxicological profile. ProtoQSAR is a five years’ company that arises from more than fifteen years of activity and experience of its team members, both academic and industrial sides, currently providing consulting services related to these predictive models used by different industries. Our goal is to make the leap from consulting through the validation and scaling up this high-potential opportunity in its sector.

TOXIFATE will help change the paradigm of chemical safety assessment and give the next generation of toxicologists the fundamental knowledge and experimental skills that are required to bring about this change. The safe use of chemicals is essential for global prosperity and human health and this is entirely dependent upon robust toxicology. For practical, scientific and ethical reasons, modern toxicology is moving to computational approaches in which large datasets describing chemical structures and toxicological outcomes are used to predict toxicity. The promise of this approach was recently demonstrated when computational approaches for the first-time outperformed animal testing in predicting human toxicity. Nonetheless, our ability to predict toxicity is poor because our understanding of how chemicals affect cells is still rudimentary. TOXIFATE will provide intersectoral training to 2 researchers who will build new assays to detect toxicant-induced changes in cells. Multi-disciplinary training will generate unique high-content and transcriptomic data describing cell stress and cell death responses. This multidimensional dataset will be used to develop novel computational approaches to improve toxicity prediction. TOXIFATE will focus on chemical-induced myotoxicity, a severe and sometimes life-threatening toxicity that is characterized by skeletal muscle breakdown. Although at least 200 drugs and chemicals cause myotoxicity, it is relatively under-investigated and what is learnt here to improve prediction of myotoxicity will also be used to improve prediction of other types of toxicity. TOXIFATE researchers will bridge the gap between state-of-the-art science and chemical hazard and risk assessment and be equipped for career paths in the emerging computational toxicology sectors. TOXIFATE will also aid the EU and the global community by quickly, ethically and economically addressing the societal challenge of better identifying chemical hazards to human health.