ELIXIR and Toxicology: a community in development

Introduction of the ELIXIR Toxicology Community

Toxicology as a field tries to understand the negative consequences that may arise from the interactions of chemicals with living organisms. In this ELIXIR (European life-sciences Infrastructure for biological Information)¹ Community, the focus will be primarily the protection of human health. There are a number of both chemical and biological “interoperability” issues key to the toxicology field that translate into data interoperability issues. These include the connection between the action and activity of a particular chemical compound to its effective amount available at a biological target (the link between toxicodynamics and toxicokinetics). Typically this is also a link between biological data analysis (including large-scale multi-omics) and kinetic modelling. Other examples include interactions between a compound and its target (a protein, nucleic sequence or membrane structure for instance). This is primarily based on the interplay between chemistry (the chemical structure and, for example, its related properties in terms of functional groups, charge, shape and related binding affinity) and biochemistry (like biomolecular 3D structures). Also, mixture toxicity needs to be considered as combinations of chemicals with synergistic or antagonistic behaviour, or a combination thereof. While chemicals with similar modes of action may act in terms of concentration addition, those with different modes may rather act according to independent action.^2,3 Substances with low toxicity may interact in concentration addition rather than as excess toxicity drivers of one compound. Often there is a need to translate the knowledge about one compound into knowledge about other compounds, where approaches like quantitative structure-activity relationships (QSARs) help to elucidate this knowledge and predict required property and toxicity, and in more general when “read-across approaches” come into play that are based on chemical data only, biological data only, or are hybrid.^4,5 These again need detailed information about the relationships between related chemical compounds, specific properties and toxicological endpoints and, when considering chemical-biological interactions, details regarding both the chemical structures and adequate descriptions of the biological targets. Since toxicological endpoints can be represented in a myriad of ways, the toxicological effect data are often scattered over multiple repositories and databases hosting different types of data; i.e. chemical structures, toxicity data (in vivo and in vitro), biological target details, and omics data. This itself is not a problem, but the separation and segregation make the data difficult to find and to connect. Currently, many of these deposition databases (where datasets can be archived) do not provide adequate descriptions regarding typical toxicological study designs and parameters, quality control, data acceptance criteria, or even clear identification of the compounds tested. This all adds to the need for an adequate FAIRification process, to make toxicology data more Findable, Accessible, Interoperable and Reusable (FAIR - see Go-FAIR).^6,7

The field of toxicology would certainly benefit from clear, standardized guidelines for data capture, approaches to integrate and connect across multiple databases, clear data licensing for these data repositories, and tools that support accessing the data (as being developed by the ELIXIR Converge data brokerage work). Existing study capture tools can be extended with templates defined for toxicology, which end up in a central place (e.g. Biosamples⁸) with links to omics and other data distributed over technology-specific deposition databases and BioStudies.⁹ Relevant portals (e.g. FAIRsharing.org^10,11) should then be able to identify these studies, also linking to existing but scattered toxicology databases.

Risk assessment, consisting of hazard identification, characterisation and risk evaluation in relation to exposure, involves expert opinions based on discussions and data interpretation. Streamlining this process is important also because of the huge number of chemicals known and produced in chemistry, biotechnology or food production, with over 350,000 chemicals documented on the global market.¹² However, expert evaluation is and will remain crucial and calls for extensive data provenance both for the actual data (how was it measured, where was it published, whom to give credit for it), and for the risk assessments themselves. Interestingly this process stumbles on problems and solutions that have much in common with other fields, where part of the data need to be hidden and other parts can be publicly accessed, as observed in pilot approaches in genetics, or patient data repositories.

Toxicology, while established as discipline, is also rapidly developing in areas, where, for example, new molecular methods to describe the adverse outcome of exposure to a toxic substance are not yet fully established. This offers opportunities for integration in systems biology approaches building on molecular pathway descriptions that benefit from modern network biology approaches. Compound profiling, where, for example, using omics methods to characterize the effect of compounds on cells, are generally useful for categorizing compounds, or for the development of predictive pathway models. For such studies, the availability of high-quality annotations for compounds is of paramount importance to enable the use of omics profiles in toxicology.

Toxicology is also an applied field. There are important applications of toxicity tests in the regulatory field and large amounts of data are collected for that purpose, often for different (governmental) agencies. Making this data more Findable and Reusable is often seen as an important way to reduce both animal testing and the cost of registration of new compounds. If some data is not available to the public due to ownership by companies or other constraints, indexes should be developed to enable the use of this data in an aggregated form (such as the SPIN index of the Swedish Chemicals agency (KEMI) and other Nordic chemical agencies). Typically, regulatory use requires very precise and rigid descriptions of protocols, including reporting methods. On the other hand, better insights into the toxicological mechanisms, including interactions between chemicals, benefit from more innovative research methods (e.g. single-cell omics, induced stem cell applications, imaging methods) and from creative development of new analysis methods. While it is beneficial to combine results from both types of approaches (termed New Approach Methods (NAM) in the regulatory field), the corresponding interoperability issues are often quite different.

User community

Europe is steadily increasing demands on the risk assessment of chemicals, drugs, cosmetics ingredients and nanomaterials to lead to safer products, resulting in a strong toxicology research community with sub-communities in, for example, the drug development, environmental, nanomaterial and rare-disease areas. Recent changes in European law for animal testing and new demands for testing of lower-volume chemicals and nanomaterials have triggered large-scale research into alternative testing approaches. These activities not only produce new biological and mechanistic insights, but also large amounts of new data, which have to be managed and shared for re-usage to avoid unnecessary duplication of experiments and, in this way, reduce animal testing. The goal is that the combination of data from integrated in vitro and in silico approaches will support (ultimately personalized) risk/benefit health analysis, safer drug innovation with fewer needs to withdraw after registration, a fact-based perception of chemical safety, safe-by-design nanomaterials and sustainable and safe economies.

The European Union supports toxicological and risk assessment projects with various funding programs. Recently, large collections of data have been released, resulting from research clusters, such as SEURAT-1,¹³ the EU NanoSafety Cluster (NSC) with NANoREG, EU-ToxRisk,¹⁴ the NORMAN Network, and the EU Innovative Medicines Initiative (IMI) funded projects related to drug toxicology, including eTOX,¹⁵ eTRANSAFE,¹⁶ and Eurion.¹⁷ Data from these and other projects are becoming available, sometimes as Open Data (e.g. NANoREG) and sometimes as FAIR data. An example of the latter is European REACH data, which has recently been made FAIR by the Cefic-LRI-funded project AMBIT-LRI.¹⁸ Furthermore, the FAIRplus project is collaborating with eTOX on making their data more FAIR, and the new Precision Toxicology will develop a data commons following the FAIR principles.

However, there are a few opportunities for data handling that need to be taken.^19,20 Recent studies show how powerful the combination of toxicology information and omics data is,^21,22 but to be able to obtain the statistical significance to draw these conclusions, data from the US and Japan had to be combined. In contrast to large data sets like DrugMatrix,²³ ToxCast/Tox21,²⁴ and TG-GATEs²⁵ from these countries, data from European projects is often not sufficiently integrated. Luckily, there are signs that the community is going in the right direction, e.g. the aforementioned data integration by diXa,²⁶ NANoREG and REACH.²⁷ With respect to the European Chemical Industry, various authors have been involved in other Cefic-LRI activities related to data management (AIMT-3, AIMT-4).

In addition, the eTOX project²⁸ has established data integration approaches e.g. to enable the development of QSARs relating chemical structures to in vivo toxicopathological outcomes. As such, the project also delivered databases and approaches to ontology development, text mining approaches, and approaches for prediction of drug metabolism and pharmacokinetic features. Moreover, in 2017 the Organisation for Economic Co-operation and Development (OECD) performed an online survey underscoring the fact that data integration for safety is of global concern for ultimate risk assessment. The purpose of the resulting knowledge base is the integration of eCHEMportal (The Global Portal to Information on Chemical Substances²⁹), IUCLID (International Uniform ChemicaL Information Database³⁰), and OECD’s QSAR Toolbox³¹ supporting the development of Adverse Outcome Pathways and associated infrastructures (AOP-Wiki).³² Despite these positive developments, the ‘data integration struggle’ from various perspectives (omics, computational chemistry and more ‘conventional’ toxicological data within REACH and pharmaceutical industry setting) remains a challenge.

Roadmap

The above initiatives are mainly driven by user communities themselves: chemical industry, funding agencies, pharmaceutical companies, governmental agencies and organisations such as Member State organisations, the OECD and non-governmental organisations (NGOs). ELIXIR can contribute strongly to the existing infrastructure projects from a cheminformatics and bioinformatics perspective, providing tools and guidelines, linking and harmonizing the ongoing activities, serving the toxicology users.

For future risk assessment paradigms solely based on human-derived models, and in this way of higher relevance for human adverse effects,³³ various data types will need to be integrated into and cross the conventional boundaries of risk assessment. This involves external exposure assessment (e.g. via workplace or environmental modelling and measurements^34,35), internal exposure characterisation (ADME-TK (absorption, distribution, metabolism, and excretion - toxicokinetics) e.g. via modelling and biomarker-based detection), toxicodynamics on a molecular level, and cell and systems biology. In this way, more data-driven mechanism-based evaluations and supporting data can integrate into regulatory risk assessment. There will not only be more but also more diverse data as e.g. internal exposure data may be inferred from biomonitoring data and/or physiologically-based toxicokinetic modelling to estimate target dose available at the active sites involved in the molecular initiating events of Adverse Outcome Pathways (AOPs).

Another relevant topic is the concept of the exposome,^36–38 which aims at characterising lifetime exposure (not only to chemicals in the narrowest sense, but also dietary components, lifestyle factors, environmental exposures, etc) in relation to health outcome. Often, vulnerable periods of life (infancy, childhood, and old age) are investigated, and the evaluation also integrates epidemiology. This is one clear demonstration of the trend that the previously distinct areas of toxicology, drug and product design and personalized/precision medicine³⁹ but also environment and health and epidemiology are moving closer together. Data sharing will be increasingly necessary across these disciplines. High throughput data analysis in exposomics, for instance, shares many parallels with metabolomics and other higher level omics analyses, with an added layer of chemical complexity on top.³⁸

The toxicology community is large and well established. The current list of proponents of an ELIXIR Toxicology Community only reflects a subset of a much larger community with a lot of commitments to, and activities around open collaboration. It has clear omics, knowledge management and data infrastructure needs to accommodate the increasing wish to predict toxicology without animal testing (e.g. in SEURAT-1, EU-ToxRisk, eTOX OBERON⁴⁰). Foreseeing this need for better infrastructures, the community has previously contacted ELIXIR for collaboration. Various domain-specific projects exist that service the toxicology community with computational and database knowledge (OpenRiskNet, NanoCommons) that can translate ELIXIR knowledge to the respective communities. These infrastructure projects are the successors of research projects focusing on data management, including diXa,²⁶ ToxBank⁴¹ and eNanoMapper.⁴²

To benefit the research community, small and medium-sized enterprises (SMEs) and larger industry, and to enable further future support to regulatory applications and upcoming calls (e.g. Green Deal^43,44), we need to reach an inclusive ecosystem of data, evaluation and modelling tools. The current separate consortia from different toxicity-related and neighbouring disciplines already work towards data and knowledge that is FAIR.⁶ After all, these aspects are essential to efficiently assess the risk of new compounds and materials, as well as combined risks of current stressors (e.g. under the exposome concept). To further accelerate these activities, more toxicology-related data and knowledge need to be linked, such as on physiologically based toxicokinetic (PBTK) modelling, biological pathways describing affected metabolism and cell biology processes, metabolism, metabolic models and metabolism predictions, drug-response, omics (biological identity), chemical structures and associated metadata (use, hazard, transformations), QSARs, AOPs, REACH dossiers, etc. Simultaneously, an extension towards the human (preclinical toxicology) discipline should be initiated, in which exposure data are combined with internal exposure and early biomarkers of effect data (e.g. from the European Human Biomonitoring Initiative (HBM4EU)⁴⁵ and environmental data from NORMAN^46,47) towards pathways of toxicity. Interoperability with the standardization efforts of clinical research by CDISC is important.

However, to reach such interoperable toxicology, resources need to be better integrated. Despite the work of many projects, their FAIR features can still be improved and applying newly developed FAIR metrics will help steer this.⁴⁸ Even though there is overlap in content with existing ELIXIR Communities (Table 1, key demands specifically fostering the integration for interoperable toxicology and risk assessment include the following roadmap^19,20):

The concrete steps forward proposed in this roadmap include:

Specifically, we will continue to grow the list of involved toxicology research groups, projects, and ELIXIR Node activities. This Community has already held a joint meeting to select the key priorities and use cases (see Tables 2 and 3), resulting in this positioning paper. The ELIXIR Toxicology Community will continue to expand and search for contributors with relevant expertise as the community and activities mature. By bootstrapping from Open Science approaches developed in aforementioned projects (e.g. Open PHACTS⁵³), the new Community will focus on mutual benefit, an open and inclusive community, solving practical community problems. The goal is not to design domain-specific approaches, but a pragmatic approach that provides FAIR and open tools from the start, allowing all toxicology and neighbouring communities to benefit from these harmonized solutions. The inclusive community will involve the existing sub-communities in pharmaceuticals, e.g. from eTOX, transQST,⁵⁴ and eTRANSAFE,¹⁶ cosmetic ingredients, e.g. from SEURAT-1, high and low-volume chemicals, e.g. from HBM4EU and soon from Horizon Europe Partnership for the Assessment of Risk from Chemicals (PARC),⁴⁷ or from different Cefic-LRI (Lang-Range Research Initiative) projects, and nanomaterials, via the NanoSafety Cluster, building on shared needs and community solutions and strongly aligned to other ELIXIR communities. Open licensing and interfaces (ontologies, standards, formats) will encourage new solutions and collaborations, which will be accessible to any organisation and every project within and outside Europe. This will allow close interoperability with toxicology communities outside Europe that also use open approaches, while at the same time allowing compatibility with closed approaches too. This dual model has been demonstrated successfully in recent projects. A prioritized roadmap is essential; the label “ELIXIR Community” would enable us to set priorities at a level above the individual projects. Existing components that will give this Community an initial boost, include, software (e.g. AMBIT with OpenTox API⁵⁵), databases (e.g. diXa, eNanoMapper, AMBIT-LRI, NORMAN-SLE, MassBank), ontologies (e.g. the eNanoMapper ontology,⁵⁶ AOP ontology⁵⁷), interoperability concepts (e.g. annotation of OpenAPIs, in collaboration with bio.tools, semantic structural searches with IDSM), teaching/education material (Bioschemas annotation of outreach activities, in collaboration with TeSS), and virtual infrastructures (e.g. OpenRiskNet Virtual Research Environment and the NORMAN Digital Sample Freezing Platform). However, each of these approaches would benefit from integration in the ELIXIR Platforms (see examples in Table 1) and with Core and National Resources. Various existing ELIXIR Communities need similar solutions, e.g. for chemical structure handling, toxicology needs proteomics and metabolomics, toxicology involves human data, and ecotoxicology has significant impact on crops and health.

The following projects have been adopting and integrating FAIR toxicology concepts, but need integration with ELIXIR Platforms and Communities: eTOX, NanoCommons (NanoSafety Cluster), EU-ToxRisk, OpenRiskNet, OpenTox Foundation, Open PHACTS Foundation, and the diXa platform. Many other projects have a specific scientific focus but also need integration and some will work on the FAIR concepts. A non-exhaustive list is ACEnano, SmartNanoTox, HeCaToS, NewGeneris, EnviroGenoMarkers, EXPOsOMICS, HELIX, ASAT,⁵⁸ PATROLS, and HEALS, as well as new projects e.g. new Horizon 2020 projects RISK-HUNT3R and HARMLESS and the new Horizon Europe project PARC. Companies and organisations will profit from this Community either as users or as providers of services on top of the infrastructure, including ECHA (FI), Edelweiss Connect (CH), IdeaConsult Ltd. (BG), Misvik Biology (FI), TNO (NL), Seven Past Nine (SI), and the Swedish Academic Consortium for Chemical Safety (SwACCS, SE). Industries showed a strong interest in toxicology, demonstrated by their activities: Cosmetics Europe was participant in the SEURAT-1 cluster; chemical industries (Nanotechnology Industries Association) are participant of the NSC; chemical branch organisation (Cefic) funds LRI projects around toxicology; pharmaceutical industries funds toxicology research via IMI projects like eTOX, eTRANSAFE, and Open PHACTS. The Research Data Alliance (RDA) organized a workshop recently about integration of toxicogenomics resources,⁵⁹ and collaboration with international organizations has already been established with, for example, the CompTox Chemicals Dashboard team of the US EPA⁶⁰ and PubChem from the US National Institutes of Health.⁶¹

Competing interests

None.

Grant information

This work received funding from the European Union’s Horizon 2020 research infrastrcuture programme via the OpenRiskNet project under grant agreement No. 731075. This project has received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement No. 681002 (EUToxRisk). This project has received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement No. 814572. SN acknowledges BMBF funding under grant number 031L0107. This work was supported by OBERON (https://oberon-4eu.com), a project funded by the European Union’s Horizon 2020 research and innovation program under the grant agreement No. 825712. This work was supported by the European Union,’s Horizon 2020 research and innovation program HBM4EU, grant agreement No. 733032 (https://www.hbm4eu.eu). Supported by the European Union’s Horizon 2020 programme under the Marie Skłodowska-Curie grant agreement No. 859891. Where authors are identified as personnel of the International Agency for Research on Cancer/World Health Organization, the authors alone are responsible for the views expressed in this article, and they do not necessarily represent the decisions, policy, or views of the International Agency for Research on Cancer/World Health Organization. This work was supported by the European Union,’s Horizon 2020 research and innovation program HARMLESS, grant agreement No. 953183. This work was supported by the European Union,’s Horizon 2020 research and innovation program Gov4Nano, grant agreement No. 814401. This work was supported by the Swedish Fund for Research Without Animal Experiments under grant number N2020-0005. This work was supported by the European Union,’s Horizon 2020 research and innovation program NTS-EXPOSURE, Grant agreement ID: 896141. This work was supported by the European Union,’s LIFE program LIFE-APEX, Grant agreement ID: LIFE17 ENV/SK/000355. This study was funded by the German Environment Agency within the PHION project (grant number 3718 674150). The Data Readiness Group is supported, in this ELIXIR Community, by the H2020 Precision Toxicology project (H2020-EU 965406). The Data Readiness Group is supported, in this ELIXIR Community, by the Wellcome ISA-InterMine project (208381/A/17/Z). The Data Readiness Group is supported, in this ELIXIR Community, by the Wellcome FAIRsharing project (212930/Z/18/Z). This work received funding from the European Union’s Horizon 2020 research and innovation programme RiskGONE Project under grant agreement No. 814425. NR’s research is funded by a Miguel Servet contract (CO19/00060) from Instituto de Salud Carlos III, cofinanced by the European Union. UM (Uni. of Tartu) is grateful for support to Ministry of Education and Research, Republic of Estonia through Estonian Research Council (grant number IUT34-14) and to European Union European Regional Development Fund through Foundation Archimedes (grant number TK143, Centre of Excellence in Molecular Cell Engineering). Development and Implementation of a Sustainable Modelling Platform for NanoInformatics. Linking LRI Ambit chemoinformatic system with the IUCLID substance database to support read-across of substance endpoint data and category formation. This work was supported by the French Ministry of Research and National Research Agency as part of the French MetaboHUB, the national metabolomics and fluxomics infrastructure (Grant ANR-INBS-0010). This project has received funding from the European Union’s Horizon 2020 research and innovation program under grant agreement GOLIATH No. 825489.

Disclaimer