396
B. Hardy et al.
algorithm, a defined domain of applicability, appropriate measures of goodness-offit, robustness and predictivity and a mechanistic interpretation [19].
Another strategy for including the in silico methods into decision making and
accelerating the regulatory acceptance for toxicity testing is represented by the Integrated Approaches to Testing and Assessment (IATAs), which are currently proposed
as practical solutions to integrate such alternative methods, with ITS being an example of such an approach. IATAs provide a means for combining the data from different
methods, considering all available relevant information about a substance in a weightof-evidence assessment, to inform regulatory hazard or risk decisions or the need for
additional tests and rely on non-animal approaches to determine chemical hazard
or risk [20]. In parallel, the development of the adverse outcome pathway (AOP)
framework, which provides information on the adverse outcome of regulatory concern, offers the biological context to facilitate development of IATAs for regulatory
decision making [21]. For example, in silico models like QSAR that are designed to
predict key events in different pathways should be useful sources of information in
IATA, whereas QSARs that are directly predictive of the adverse effect may be useful
for increasing confidence in weight-of-evidence arguments [16]. However, a critical
question for the risk assessor when applying a non-testing method for regulatory
purposes is regarding the reproducibility of the model, thus on the reliability of the
prediction.
We have selected OpenTox which is the leading global open knowledge community in new in silico toxicology methods as the ideal location for discussion of
case studies and the above best practices of potential to benefit emerging regulatory
frameworks. We aim to demonstrate with a broad community input on how new
technologies are essential for regulatory science, more specifically by highlighting
reproducible in silico practice via OpenTox, with a focus theme on the issues of reproducibility, predictive modeling, and other related enabling topics such as semantic
interoperability of contributing resources. The process is not only about building
predictive models, but also about placing observations on how predictive uses are
constantly changing within a community evaluation context. In addition to building an application with a set of principles, other concerns shared by developers and
practitioners are the implementation of best practices based on quality, reliability,
robustness, interoperability, reproducibility, harmonization, completeness, openness,
and confidence. These principles and their associated practices require both a dialogue and a consensus such as best practice protocols, context of use and fit for
purpose issues. Another current related initiative is the in silico toxicology (IST)
protocol consortium, organized by Glenn Myatt, Founder of Leadscope. This international consortium includes regulators, government agencies, industry, academics,
model developers, and consultants across many different sectors, formed initially
with the intention of creating the overall strategy for an in silico protocol development. Working subgroups will develop individual in silico toxicology protocols
for major toxicological endpoints, including genetic toxicity, carcinogenicity, acute
toxicity, reproductive toxicity, and developmental toxicity, and we plan to interact
with this important initiative.
B. Hardy et al.
algorithm, a defined domain of applicability, appropriate measures of goodness-offit, robustness and predictivity and a mechanistic interpretation [19].
Another strategy for including the in silico methods into decision making and
accelerating the regulatory acceptance for toxicity testing is represented by the Integrated Approaches to Testing and Assessment (IATAs), which are currently proposed
as practical solutions to integrate such alternative methods, with ITS being an example of such an approach. IATAs provide a means for combining the data from different
methods, considering all available relevant information about a substance in a weightof-evidence assessment, to inform regulatory hazard or risk decisions or the need for
additional tests and rely on non-animal approaches to determine chemical hazard
or risk [20]. In parallel, the development of the adverse outcome pathway (AOP)
framework, which provides information on the adverse outcome of regulatory concern, offers the biological context to facilitate development of IATAs for regulatory
decision making [21]. For example, in silico models like QSAR that are designed to
predict key events in different pathways should be useful sources of information in
IATA, whereas QSARs that are directly predictive of the adverse effect may be useful
for increasing confidence in weight-of-evidence arguments [16]. However, a critical
question for the risk assessor when applying a non-testing method for regulatory
purposes is regarding the reproducibility of the model, thus on the reliability of the
prediction.
We have selected OpenTox which is the leading global open knowledge community in new in silico toxicology methods as the ideal location for discussion of
case studies and the above best practices of potential to benefit emerging regulatory
frameworks. We aim to demonstrate with a broad community input on how new
technologies are essential for regulatory science, more specifically by highlighting
reproducible in silico practice via OpenTox, with a focus theme on the issues of reproducibility, predictive modeling, and other related enabling topics such as semantic
interoperability of contributing resources. The process is not only about building
predictive models, but also about placing observations on how predictive uses are
constantly changing within a community evaluation context. In addition to building an application with a set of principles, other concerns shared by developers and
practitioners are the implementation of best practices based on quality, reliability,
robustness, interoperability, reproducibility, harmonization, completeness, openness,
and confidence. These principles and their associated practices require both a dialogue and a consensus such as best practice protocols, context of use and fit for
purpose issues. Another current related initiative is the in silico toxicology (IST)
protocol consortium, organized by Glenn Myatt, Founder of Leadscope. This international consortium includes regulators, government agencies, industry, academics,
model developers, and consultants across many different sectors, formed initially
with the intention of creating the overall strategy for an in silico protocol development. Working subgroups will develop individual in silico toxicology protocols
for major toxicological endpoints, including genetic toxicity, carcinogenicity, acute
toxicity, reproductive toxicity, and developmental toxicity, and we plan to interact
with this important initiative.