PurposeThe purpose of this Activity is to demonstrate your understanding of the concepts learned in this week’s readings/ educational videos.Action Items
SPECIALTY GRAND CHALLENGE
published: 25 June 2020
doi: 10.3389/ftox.2020.00002
Navigating the Minefield of
Computational Toxicology and
Informatics: Looking Back and
Charting a New Horizon
Grace Patlewicz*
Independent Researcher, Durham, NC, United States
Keywords: informatics, read-across, data science, computational toxicology, (Q)SAR
INTRODUCTION
Edited by:
Ruili Huang,
National Center for Advancing
Translational Sciences (NCATS),
United States
Reviewed by:
Zhichao Liu,
National Center for Toxicological
Research (FDA), United States
Hao Zhu,
Rutgers, The State University of New
Jersey, United States
*Correspondence:
Grace Patlewicz
patlewig@hotmail.com
Specialty section:
This article was submitted to
Computational Toxicology and
Informatics,
a section of the journal
Frontiers in Toxicology
Received: 09 March 2020
Accepted: 20 May 2020
Published: 25 June 2020
Citation:
Patlewicz G (2020) Navigating the
Minefield of Computational Toxicology
and Informatics: Looking Back and
Charting a New Horizon.
Front. Toxicol. 2:2.
doi: 10.3389/ftox.2020.00002
Frontiers in Toxicology | www.frontiersin.org
As we enter 2020, it is worth looking back at the development and progression of the computational
toxicology discipline, how it has evolved and what some opportunities might be going forward.
Computational toxicology stands poised to broadly and directly inform chemical safety assessment,
and as such, the demands of computational toxicology are growing due to international regulatory
needs. Critical to increasing scientific confidence in the use of computational toxicology approaches
in applied toxicology decision-making will be: (1) transparency and reproducibility in the
underlying data and data analysis approaches utilized; (2) accessibility of information to evaluate
the fitness of the computational toxicology approach for a particular problem; and (3) sharing of
ideas and approaches internationally. Herein the progress in applied computational toxicology is
considered, with a call for additional research to continue this rapid advancement.
EARLY COMPUTATIONAL TOXICOLOGY: APPLICATION OF
QUANTITATIVE STRUCTURE ACTIVITY RELATIONSHIPS
[(Q)SARs]
A quarter of century ago, the field of computational toxicology might simply have been
summarized as the intersection of three scientific domains: toxicology, chemistry, and statistics,
packaged in predictive models such as SAR and QSAR models, collectively referred to as
(Quantitative) Structure Activity Relationships [(Q)SARs]. (Q)SARs are theoretical models that
can be used to predict in a quantitative (e.g., potency) or qualitative manner (e.g., active/inactive)
the physicochemical, biological [e.g., an (eco)toxicological endpoint] and environmental fate
properties of compounds from the knowledge of their chemical structure (Worth et al., 2005). A
SAR is a (qualitative) association between a chemical substructure and the potential of a chemical
containing the substructure to exhibit a certain biological effect. The classic example of a SAR
was the supramolecule published by Ashby and Tennant (1988) which related chemical structure
moieties to genotoxic carcinogenicity. Typical toxicity endpoints under study were those with a
greater preponderance of data such as the Ames test for bacterial mutagenicity (see Benigni and
Bossa, 2019 for a recent review), or the fathead minnow fish acute lethality test (Adhikari and
Mishra, 2018). Physicochemical properties such water solubility, octanol-water partition coefficient
(LogKow) were also modeled. The algorithms underpinning these predictive models (QSAR)
tended not to be overly complex mainly because the data volume was usually limited, “big” data was
confined to a few hundred data points and typically far less. The algorithms used to develop QSARs
relied upon conventional statistical approaches such as linear regression or logistic regression,
1
June 2020 | Volume 2 | Article 2
Patlewicz
Computational Toxicology: A New Horizon
validation namely: a defined endpoint, unambiguous algorithm,
appropriate measures of predictivity (e.g., external validation),
goodness of fit (e.g., cross-validation), an applicability domain
and mechanistic interpretation if possible (OECD, 2004, 2007;
Patlewicz et al., 2016). The QSAR Validation principles largely
provided the impetus to develop new approaches to characterize
the applicability domain of models (Netzeva et al., 2005;
Nikolova-Jeliazkova and Jaworska, 2005) as well as consider
integration of models e.g., consensus models (e.g., Votano
et al., 2004). In the development of Frontiers in Toxicology:
Computational Toxicology and Informatics, a focus on the
validation principles and their applicability to (Q)SARs and
beyond is a component of advancing the scientific confidence in
using these approaches in applied decision-making.
in part because the data volume did not merit more complex
models, and in part since the models being developed relied
on a limited set of descriptors that could be readily computed
and interpreted relative to the property being modeled. Indeed,
many fish toxicity models and Ames models relied upon LogKow
as the main determining factor. Computing descriptors for
chemicals was mainly dependent on commercial software within
specific QSAR modeling platforms, e.g., TSAR from Oxford
Molecular, Biovia’s QSAR Workbench (https://www.3dsbiovia.
com/products/collaborative-science/biovia-qsar-workbench/).
The types of QSAR models for toxicity were often “local,” i.e.,
based on defined mechanisms or chemical classes. The exception
tended to be for physicochemical parameters where models
were categorized as “global,” i.e., heterogenous training datasets
comprising a diversity in chemical structure.
The underlying principle within this framework was that
the toxicity (property) being predicted was a function of
chemical structure. The notion of similarity where similar
chemicals were expected to cause similar toxicities also formed
the complementary basis around the concept of read-across
(Patlewicz et al., 2018) as well as Thresholds for Toxicological
Concern (TTC) approaches (Kroes et al., 2004).
The application of these models was also limited, usually in
providing preliminary indications of activity rather than in lieu
of additional empirical data. The toxicity would be characterized
by a single summary endpoint e.g., point of departure such as
a No Adverse Effect Level (Concentration) [NOAEL(C)] and
usually a single value for a given substance. The concept of
reproducibility of the test method was not a major consideration,
since studies tended not be repeated due to cost, animal use, and
time constraints.
BROADENING COMPUTATIONAL
TOXICOLOGY TO A STRATEGIC IN SILICO
AND IN VITRO APPROACH AS
SUPPORTED BY INFORMATICS
At the same time, the NRC report was published (NRC, 2007)
which outlined the change in how toxicity testing could be
undertaken. Subsequent reports on computational methodology
for exposure (NRC, 2012) and risk assessment (NRC, 2017) have
broadened the call. The NRC reports, together with the synergism
of increased computing resources, increased access to laboratory
automation for toxicology, and development of methodologies
that efficiently generated large volumes of data, generated
a disruptive change in the field and an expansion of what
computational toxicology represented. Instead of summarizing
toxicity on the basis of traditional toxicity tests, a shift was
proposed to predict genotoxic vs. non-genotoxic substances,
and then to have in vitro bioactivity and predicted exposure
define a bioactivity:exposure ratio, which would inform the
need for models of greater biological complexity (Thomas
et al., 2013, 2019). This shift is dependent on high throughput
and high content screening methods (HTS/HCS), including
high throughput transcriptomics (HTTr) and high throughput
phenotypic profiling (HTTP) of cellular morphology (Harrill
et al., 2019; Thomas et al., 2019; Nyffeler et al., 2020).
The data needed for a rapid, high-throughput safety
assessment requires application of a range of computational
approaches for data analysis, data storage, and in silico predictive
modeling. These challenges are directly identified in the title
of this journal as the necessary “informatics” component of
realizing computational toxicology for safety assessment. How
to meet these informatic challenges is the subject of ongoing
research as the volume and variety of data require tools for
large scale data processing, databasing and informatics for singledimension and multi-dimensional datasets, visualization for
heterogeneous information, demonstrating reproducibility, and
quality control, and perhaps most challenging, for interpretation
and communication in the appropriate format and context for
chemical safety assessment. Many aspects of the vision articulated
by the initial NRC report have been realized in preliminary
form by the ToxCast (Kavlock et al., 2012) and Tox21 research
EVOLVING REGULATORY LANDSCAPE
FOR (Q)SARs IN APPLICATION
In the late 1990s, there started to be a need to make predictions
for a broader coverage of chemicals beyond the smaller datasets
that underpinned the mechanistic chemical class type QSAR
models to date. The shift was partly driven by interest in
improvements to quantitative descriptions of chemical structure
for toxicity prediction, and the availability of computing power.
Decision contexts were also changing and provided an
additional impetus for new model development. Two main
drivers were influencing this change: the need for non-animal
alternatives, largely prompted the EU Cosmetics Regulation
(European Commission, 2009) and the EU Chemicals legislation
known as REACH (European Commission, 2006). REACH in
particular had a profound effect in the development, evaluation,
and application of QSARs, primarily since the decision context
was to use QSAR predictions as supporting information in the
construct of an Integrated Approaches to Testing and Assessment
(IATA) (Tollefsen et al., 2014) and/or in lieu of new experimental
testing. In the run up to REACH coming into force, there was a
concerted effort to characterize a framework to facilitate the use
of (Q)SARs for regulatory purposes (Cronin et al., 2003a,b). This
culminated in the formulation of the OECD QSAR principles for
Frontiers in Toxicology | www.frontiersin.org
2
June 2020 | Volume 2 | Article 2
Patlewicz
Computational Toxicology: A New Horizon
by open source libraries developed on top of programming
languages such as R and python. A skill set that so far has not
been a strong focus as yet is that of data engineer, the backend
of data science: models developed need to be deployed, and for
reproducible models, a different set of considerations regarding
versioning of models, their underlying inputs, and algorithms
e.g., docker containers.
programs (Tice et al., 2013; Thomas et al., 2018) which have
generated publicly available HTS data on thousands of chemicals.
In addition to the data generated, data processing pipelines have
been developed (Hsieh et al., 2015; Filer et al., 2017) and many
different models continue to be derived using the data, including
those designed to understand mode-of-action (e.g., Shah et al.,
2011; Judson et al., 2015; Kleinstreuer et al., 2016; Saili et al.,
2019;) as well as models that use HTS data as descriptors or
training information for (Q)SARs (e.g., Liu et al., 2015; Mansouri
et al., 2016). The informatic needs of data driven predictive
modeling, and how to standardize and openly transmit these
models, is a clear need in the field. Recent progress in advancing
computational toxicology and the associated challenges were
discussed in Ciallela and Zhu (2019). Noteworthy examples of
recent data driven models include those for acute oral toxicity
(Russo et al., 2019) and liver toxicity (Zhao et al., 2020).
The databasing and informatic challenges for computational
toxicology also have a legacy component: to bolster scientific
confidence and for fit-for-purpose evaluations, the use of in
vivo animal study data and any available in vivo human data
have been important in the early application of computational
toxicology as replacements or alternatives to existing approaches
(Kleinstreuer et al., 2016, uterotrophic database; Hoffmann et al.,
2018, LLNA database; Watford et al., 2019b; ToxRefDBv2). To
enable quantitative comparisons of dose in animals or humans,
internal exposures as modeled using HT toxicokinetic data and
modeling (Wetmore et al., 2012; Pearce et al., 2017; Wambaugh
et al., 2018) have been developed to support in vitro to in vivo
extrapolation (IVIVE). Examples of how IVIVE has enabled
greater utilization of HTS data for safety assessment are discussed
in more detail in Thomas et al. (2019), Paul Friedman et al.
(2020).
Evaluating Fit-For-Purpose Utility
The variety and volume of data now being generated and
analyzed has also raised challenging questions for the “legacy”
or existing in vivo data available: the level of curation, study
reproducibility, and how these data may be used to benchmark
new approach methodologies are all of high interest (Pham et al.,
2019) Using in vivo study data to benchmark the performance
of or directly train new approach methodologies for human or
ecological health assessment should include some evaluation of
how variable the in vivo study data may have been. Fit-forpurpose evaluations require not only acquisition of and curation
of reference data and meaningful assessments of variability and
uncertainty, but also efforts to increase data interoperability
(Watford et al., 2019a). That requires ontologies in data storage
and domain knowledge, as well as standards that permit sharing
and exchange of data and models. Another consideration is the
fact that these new data stream technologies are evergreen and in
constant state of evolution and improvement. Evaluation of the
fitness of these information needs to be flexible to deal with the
changes in the methods of a specific technology and increased
understanding of method performance using a large number of
substances (Judson et al., 2018; Ciallela and Zhu, 2019).
The concept of an “applicability domain” is central in an
evaluation of fit-for-purpose use of computational toxicology
approaches, taking on an extended meaning as we want to
understand the relevance of the model and when it can be
applied and the extent to which it can be used to forecast
other substances and to what extent there is confidence for
that to occur. The uncertainties associated with the prediction
need to be clearly specified and linked back to the decision
context and purpose intended. The appropriate measures of
fit and predictivity remain important considerations. What are
the steps and procedures that were applied during the model
building phases, including selection of approach, cross validation,
performance metrics, and hyperparameter optimization. Before
final model evaluation and application to new data (prediction),
consideration of how to deploy a model should include a plan for
ensuring reproducibility.
CURRENT STATE OF COMPUTATIONAL
TOXICOLOGY AND INFORMATICS
The Rising Need of Informatics and Data
Engineering
The validation principles of QSARs are perhaps still relevant
today but in providing a framework to make explicit the
provenance of the data, how it has been processed, the
assumptions made, and the transparency and reproducibility of
any models derived (Patlewicz et al., 2015). The three pillars
of statistics, toxicology, and chemistry have since extended,
in part due to the demand to make rapid decisions (Judson
et al., 2010), with greater transparency. Perhaps the term “Data
Science” now better captures the skill sets and needs encompassed
in computational toxicology. Thus, computational toxicology
considers the disciplines of toxicology, chemistry, and statistics,
but also a number of front-end data science techniques relying on
programming skills to facilitate data acquisition, data processing,
storage and retrieval, data manipulation and interpretation, and
beyond traditional statistics, other machine learning and deep
learning techniques. The wealth of open source tools has also
facilitated the change in the skills and approaches now applied.
Commercial bespoke tools are being somewhat superseded
Frontiers in Toxicology | www.frontiersin.org
And Now: A Call for Research and Action
Using the QSAR Validation Principles as a
Guide
Clearly the landscape of computational toxicology has
significantly evolved in the last two decades and much
progress has already been made. Notable examples include data
challenges that have been organized by NCATS (e.g., Huang and
Xia, 2017) and NICEATM (e.g. Kleinstreuer et al., 2018). Most
progress has been realized for individual discrete substances
3
June 2020 | Volume 2 | Article 2
Patlewicz
Computational Toxicology: A New Horizon
but major areas of effort that remain to be tackled include: (1)
the challenge of big data itself e.g., how to fit for large datasets
(Ciallela and Zhu, 2019) (2) difficult substances to test in existing
HTS systems, e.g., volatiles, insoluble in solvents; (3) mixtures–to
date progress has been made on developing individual models
but less focus on ensemble models; (4) wide implementation of
cloud resources for data accessibility and data processing; (5)
metabolism and degradation aspects in inferring effects of parent
chemicals; (6) the use of unbalanced datasets in model training
and development; (7) predicting dose response in conjunction
with effects rather than extracting a summary metric from a
study (Moran et al., 2019); (8) mining and extraction of insights
from unstructured literature data; (9) standardized application
of epidemiology; and likely a myriad of other challenges yet to
be identified.
In many respects the (Q)SAR validation principles from
2004 remain relevant. The defined endpoint of a model or new
approach methodology, the purpose and the goal of the model,
and its basis need to be specified, albeit characterized differently
to meet the requirements of 2020 and beyond. “The unambiguous
nature of the algorithm” reads now as a call for increased
reproducibility of the methodology and the approach. Are the
assumptions of the modeling approach and the underlying data
clearly specified? What are the data processing steps taken?
How has the data been generated, summarized, and stored?
These and other considerations should feature prominently in
Computational Toxicology and Informatics.
REFERENCES
chemical screening data in regulatory decisions. Curr. Opin. Toxicol. 15, 64–75.
doi: 10.1016/j.cotox.2019.05.004
Hoffmann, S., Kleinstreuer, N., Alépée, N., Allen, D., Api, A. M., Ashikaga,
T., et al. (2018). Non-animal methods to predict skin sensitization
(I): the Cosmetics Europe database. Crit. Rev. Toxicol. 48, 344–358.
doi: 10.1080/10408444.2018.1429385
Hsieh, J. H., Sedykh, A., Huang, R., Xia, M., and Tice, R. R. (2015). A
data analysis pipeline accounting for artifacts in Tox21 quantitative
high-throughput screening assays. J. Biomol. Screen. 20, 887–897.
doi: 10.1177/1087057115581317
Huang, R., and Xia, M. (2017). Editorial: Tox21 Challenge to build predictive
models of nuclear receptor and stress response pathways as mediated by
exposure to environmental toxicants and drugs. Front. Environ. Sci. 5:3
doi: 10.3389/fenvs.2017.00003
Judson, R. S., Magpantay, F. M., Chickarmane, V., Haskell, C., Tania, N., Taylor, J.,
et al. (2015). Integrated model of chemical perturbations of a biological pathway
using 18 in vitro high-throughput screening assays for the estrogen receptor.
Toxicol. Sci. 148, 137–154. doi: 10.1093/toxsci/kfv168
Judson, R. S., Martin, M. T., Reif, D. M., Houck, K. A., Knudsen, T. B., Rotroff,
D. M., et al. (2010). Analysis of eight oil spill dispersants using rapid, in vitro
tests for endocrine and other biological activity. Environ. Sci. Technol. 44,
5979–5985. doi: 10.1021/es102150z
Judson, R. S., Thomas, R. S., Baker, N., Simha, A., Howey, X. M., Marable, C., et al.
(2018). Workflow for defining reference chemicals for assessing performance
of in vitro assays. ALTEX 36, 261–276. doi: 10.14573/altex.1809281
Kavlock, R., Chandler, K., Houck, K., Hunter, S., Judson, R., Kleinstreuer, N.,
et al. (2012). Update on EPA’s ToxCast program: providing high throughput
decision support tools for chemical risk management. Chem. Res. Toxicol. 25,
1287–1302. doi: 10.1021/tx3000939
Kleinstreuer, N. C., Ceger, P. C., Allen, D. G., Strickland, J., Chang, X., Hamm, J.
T., et al. (2016). A curated database of rodent uterotrophic bioactivity. Environ.
Health Perspect. 124, 556–562. doi: 10.1289/ehp.1510183
Kleinstreuer, N. C., Karmaus, A., Mansouri, K., Allen, D. G., Fitzpatrick, J. M.,
et al. (2018). Predictive models for acute oral systemic toxicity: a workshop
to bridge the gap from research to regulation. Comput. Toxicol. 8, 21–24.
doi: 10.1016/j.comtox.2018.08.002
Kroes, R., Renwick, A. G., Cheeseman, M., Kleiner, J., Mangelsdorf, I., Piersma,
A., et al. (2004). Structure-based thresholds of toxicological concern (TTC):
guidance for application to substances present at low levels in the diet. Food
Chem. Toxicol. 42, 65–83. doi: 10.1016/j.fct.2003.08.006
Liu, J., Mansouri, K., Judson, R. S., Martin, M. T., Hong, H., Chen, M., et al. (2015).
Predicting hepatotoxicity using ToxCast in vitro bioactivity and chemical
structure. Chem. Res. Toxicol. 28, 738–751. doi: 10.1021/tx500501h
AUTHOR CONTRIBUTIONS
GP prepared and wrote this article.
ACKNOWLEDGMENTS
GP thanks K. Paul Friedman for insightful comments
and discussion.
Adhikari, C., and Mishra, B. K. (2018). Quantitative structure-activity relationships
of aquatic narcosis: a review. Curr. Comput. Aided Drug Des. 14, 7–28.
doi: 10.2174/1573409913666170711130304
Ashby, J., and Tennant, R. W. (1988). Chemical structure, Salmonella mutagenicity
and extent of carcinogenicity as indicators of genotoxic carcinogenesis among
222 chemicals tested in rodents by the U.S. NCI/NTP. Mutat. Res. 204, 17–115.
doi: 10.1016/0165-1218(88)90114-0
Benigni, R., and Bossa, C. (2019). Data-based review of QSARs for
predicting genotoxicity: the state of the art. Mutagenesis 34, 17–23.
doi: 10.1093/mutage/gey028
Ciallela, H. L., and Zhu, H. (2019). Advancing computational toxicology in
the Big data era by artificial intelligence: data-driven and mechanismdriven modelling for chemical toxicity. Chem. Res. Toxicol. 32, 536–547
doi: 10.1021/acs.chemrestox.8b00393
Cronin, M. T., Jaworska, J. S., Walker, J. D., Comber, M. H., Watts, C. D.,
and Worth, A. P. (2003a). Use of QSARs in international decision-making
frameworks to predict health effects of chemical substances. Environ. Health
Perspect. 111, 1391–1401. doi: 10.1289/ehp.5760
Cronin, M. T., Walker, J. D., Jaworska, J. S., Comber, M. H., Watts, C.
D., and Worth, A. P. (2003b). Use of QSARs in international decisionmaking frameworks to predict ecologic effects and environmental fate of
chemical substances. Environ. Health Perspect. 111, 1376–1390. doi: 10.1289/
ehp.5759
European Commission (2006). Regulation (EC) No 1907/2006 of the European
Parliament and of the Council of 18 December 2006 concerning the
Registration, Evaluation, Authorisation and Restriction of Chemicals
(REACH), establishing a European Chemicals Agency, amending Directive
1999/45/EC and repealing Council Regulation (EEC) No 793/93 and
Commission Regulation (EC) No 1488/94 as well as Council Directive
76/769/EEC and Commission Directives 91/155/EEC, 93/67/EEC, 93/105/EC
and 2000/21/EC. Off. J. Eur. Union L136:3. Available online at: http://data.
europa.eu/eli/reg/2006/1907/2014-04-10
European Commission (2009). Regulation (EC) No 1223/2009 of the European
Parliament and the Council of 30 November 2009 on cosmetic products. Off.
J. Eur. Union L342:59. Available online at: http://data.europa.eu/eli/reg/2009/
1223/oj
Filer, D. L., Kothiya, P., Setzer, R. W., Judson, R. S., and Martin, M. T. (2017).
tcpl: the ToxCast pipeline for high-throughput screening data. Bioinformatics
33:618. doi: 10.1093/bioinformatics/btw680
Harrill, J., Shah, I., Setzer, R. W., Haggard, D., Auerbach, S., Judson, R., et al.
(2019). Considerations for strategic use of high-throughput transcriptomics
Frontiers in Toxicology | www.frontiersin.org
4
June 2020 | Volume 2 | Article 2
Patlewicz
Computational Toxicology: A New Horizon
Thomas, R. S., Bahadori, T., Buckley, T. J., Cowden, J., Deisenroth, C., Dionisio,
K. L., et al. (2019). The next generation blueprint of computational toxicology
at the U.S. Environmental Protection Agency. Toxicol. Sci. 169, 317–332.
doi: 10.1093/toxsci/kfz058
Thomas, R. S., Paules, R. S., Simeonov, A., Fitzpatrick, S. C., Crofton, K. M., Casey,
W. M., et al. (2018). The US Federal Tox21 Program: a strategic and operational
plan for continued leadership. ALTEX 35, 163–168. doi: 10.14573/altex.
1803011
Thomas, R. S., Philbert, M. A., Auerbach, S. S., Wetmore, B. A., Devito, M.
J., Cote, I., et al. (2013). Incorporating new technologies into toxicity
testing and risk assessment: moving from 21st century vision to a
data-driven framework. Toxicol. Sci. 136, 4–18. doi: 10.1093/toxsci/
kft178
Tice, R. R., Austin, C. P., Kavlock, R. J., and Bucher, J. R. (2013). Improving the
human hazard characterization of chemicals: a Tox21 update. Environ. Health
Perspect. 121, 756–765. doi: 10.1289/ehp.1205784
Tollefsen, K. E., Scholz, S., Cronin, M. T., Edwards, S. W., de Knecht,
J., Crofton, K., et al. (2014). Applying Adverse Outcome Pathways
(AOPs) to support Integrated Approaches to Testing and Assessment
(IATA). Regul. Toxicol. Pharmacol. 70, 629–640. doi: 10.1016/j.yrtph.2014.
09.009
Votano, J. R., Parham, M., Hall, L. H., Keir, L. B., Oloff, S., et al. (2004). Three new
consensus QSAR models for the prediction of Ames genotoxicity. Mutagenesis
19, 365–377. doi: 10.1093/mutage/geh043
Wambaugh, J. F., Hughes, M. F., Ring, C. L., MacMillan, D. K., Ford, J., Fennell,
T. R., et al. (2018). Evaluating in vitro-in vivo extrapolation of toxicokinetics.
Toxicol. Sci. 163, 152–169. doi: 10.1093/toxsci/kfy020
Watford, S., Edwards, S., Angrish, M., Judson, R. S., and Paul Friedman, K.
(2019a). Progress in data interoperability to support computational toxicology
and chemical safety evaluation. Toxicol. Appl. Pharmacol. 380:114707.
doi: 10.1016/j.taap.2019.114707
Watford, S., Ly Pham, L., Wignall, J., Shin, R., Martin, M. T., and
Friedman, K. P. (2019b). ToxRefDB version 2.0: improved utility for
predictive and retrospective toxicology analyses. Reprod. Toxicol. 89, 145–158.
doi: 10.1016/j.reprotox.2019.07.012
Wetmore, B. A., Wambaugh, J. F., Ferguson, S. S., Sochaski, M. A., Rotroff, D.
M., et al. (2012). Integration of dosimetry, exposure, and high-throughput
screening data in chemical toxicity assessment. Toxicol. Sci. 125, 157–174.
doi: 10.1093/toxsci/kfr254
Worth, A. P., Bassan, A., Gallegos, A., Netzetva, T. I., Patlewicz, G.,
Pavan, M., et al. (2005). The characterisation of (Q)uantitative StructureActivity Relationships: preliminary guidance. EUR 21866EN. Available online
at: https://publications.jrc.ec.europa.eu/repository/bitstream/JRC31241/QSAR
%20characterisation_EUR%2021866%20EN.pdf (accessed June 14, 2020).
Zhao, L., Russo, D. P., Wang, W., Aleksunes, L. M., and Zhu, H. (2020).
Mechanism-Driven Read-across of chemical hepatotoxicants based
on chemical structures and biological data. Toxicol. Sci. 174, 178–188.
doi: 10.1093/toxsci/kfaa005
Mansouri, K., Abdelaziz, A., Rybacka, A., Roncaglioni, A., Tropsha, A., Varnek,
A., et al. (2016). CERAPP: collaborative estrogen receptor activity prediction
project. Environ. Health Perspect. 124, 1023–1033. doi: 10.1289/ehp.1510267
Moran, K. R., Dunson, D., and Herring, A. H. (2019). Bayesian joint modeling of
chemical structure and dose response curves. arXiv arXiv:1912.12228. Available
online at: https://arxiv.org/abs/1912.12228
Netzeva, T. I., Worth, A., Aldenberg, T., Benigni, R., Cronin, M. T.,
et al. (2005). Current status of methods for defining the applicability
domain of (quantitative) structure-activity relationships. The report and
recommendations of ECVAM Workshop 52. Altern. Lab. Anim. 33, 155–173.
doi: 10.1177/026119290503300209
Nikolova-Jeliazkova, N., and Jaworska, J. (2005). An approach to determining
applicability domains for QSAR group contribution models: an analysis of SRC
KOWWIN. Altern. Lab. Anim. 33, 461–470. doi: 10.1177/026119290503300510
NRC (2007). Toxicity Testing in the 21st Century: A Vision and a Strategy.
Washington DC: National Academy Press.
NRC (2012). Exposure Science in the 21st Century: A Vision and A Strategy.
Washington DC: National Academy Press.
NRC (2017). Using 21st Century Science to Improve Risk-Related Evaluations.
Washington DC: National Academy Press.
Nyffeler, J., Willis, C., Lougee, R., Richard, A., Paul-Friedman, K., and Harrill, J. A.
(2020). Bioactivity screening of environmental chemicals using imaging-based
high-throughput phenotypic profiling. Toxicol. Appl. Pharmacol. 389:114876.
doi: 10.1016/j.taap.2019.114876
OECD (2004). ENV/JM/MONO/(2004)24. Available online at: http://www.oecd.
org/env/ehs/risk-assessment/37849783.pdf (accessed June 14, 2020).
OECD (2007). Guidance Document on the Validation of (Quantitative) StructureActivity Relationships [(Q)SAR] Models. Series on Testing and Assessment No.
69. OECD Environment Health and Safety Publications.
Patlewicz, G., Cronin, M. T. D., Helman, G., Lambert, J. C., Lizarraga,
L., and Shah, I. (2018). Navigating through the minefield of readacross frameworks: a commentary perspective. Comput. Toxicol. 6, 39–54.
doi: 10.1016/j.comtox.2018.04.002
Patlewicz, G., Simon, T. W., Rowlands, J. C., Budinsky, R. A., and Becker,
R. A. (2015). Proposing a scientific confidence framework to help support
the application of adverse outcome pathways for regulatory purposes. Regul.
Toxicol. Pharmacol. 71, 463–477. doi: 10.1016/j.yrtph.2015.02.011
Patlewicz, G., Worth, A. P., and Ball, N. (2016). Validation of Computational
Methods. Adv. Exp. Med. Biol. 856:165. doi: 10.1007/978-3-319-33826-2_6
Paul Friedman, K., Gagne, M., Loo, L. H., Karamertzanis, P., Netzeva, T., Sobanski,
T., et al. (2020). Utility of in vitro bioactivity as a lower bound estimate of in vivo
adverse effect levels and in risk-based prioritization. Toxicol. Sci. 173, 202–225.
doi: 10.1093/toxsci/kfz201
Pearce, R. G., Setzer, R. W., Strope, C. L., Wambaugh, J. F., and Sipes, N. S. (2017).
httk: R package for high-throughput toxicokinetics. J. Stat. Softw. 79, 1–26.
doi: 10.18637/jss.v079.i04
Pham, L., Sheffield, T. Y., Pradeep, P., Brown, J., Haggard, D. E., Wambaugh,
J., et al. (2019). Estimating uncertainty in the context of new approach
methodologies for potential use in chemical safety evaluation. Curr. Opin.
Toxicol. 15, 40–47. doi: 10.1016/j.cotox.2019.04.001
Russo, D. P., Strickland, J., Karmaus, A. L., Wang, W., Shende, S., et al. (2019). Non
animal models for acute toxicity evaluations: applying data-driven profiling and
read-across. Environ. Health Perspect. 127:47001. doi: 10.1289/EHP3614
Saili, K. S., Franzosa, J. A., Baker, N. C., Ellis-Hutchings, R. G., Settivari, R. S.,
Carney, E. W., et al. (2019). Systems modeling of developmental vascular
toxicity. Curr. Opin. Toxicol. 15, 55–63. doi: 10.1016/j.cotox.2019.04.004
Shah, I., Houck, K., Judson, R. S., Kavlock, R. J., Martin, M. T., Reif, D. M., et al.
(2011). Using nuclear receptor activity to stratify hepatocarcinogens. PLoS ONE
6:e14584. doi: 10.1371/journal.pone.0014584
Frontiers in Toxicology | www.frontiersin.org
Conflict of Interest: The author declares that the research was conducted in the
absence of any commercial or financial relationships that could be construed as a
potential conflict of interest.
Copyright © 2020 Patlewicz. This is an open-access article distributed under the
terms of the Creative Commons Attribution License (CC BY). The use, distribution
or reproduction in other forums is permitted, provided the original author(s) and
the copyright owner(s) are credited and that the original publication in this journal
is cited, in accordance with accepted academic practice. No use, distribution or
reproduction is permitted which does not comply with these terms.
5
June 2020 | Volume 2 | Article 2
Order an Essay Now & Get These Features For Free:
Turnitin Report
Formatting
Title Page
Citation
Outline
