Chemical Universe for Endocrine Disruptor Screening Program (EDSP) and Guiding Validation Standards
The United States Environmental Protection Agency (EPA) has released a document titled "Endocrine Disruptor Screening Program Universe of Chemicals and General Validation Principles." This document aims to guide staff and managers within the EPA, providing principles for the validation of computational toxicology tools in the context of the Endocrine Disruptor Screening Program (EDSP).
The focus of the document is on the EDSP universe of chemicals and computational toxicology. It outlines stringent validation criteria for computational toxicology tools, ensuring their reliability, transparency, and applicability to the relevant chemical domain.
The principles include a clear definition of the chemical domain of applicability, specifying which chemicals the model or tool is suitable for. Data source transparency and verification are also key, ensuring that validation data are well documented, curated, and discrepancies between original study reports and databases like EPA’s ToxVal DB are examined and explained.
The use of well-curated toxicological databases for tool training and validation is another important principle, with explicit versioning and date-of-collection information to maintain reproducibility and track updates. Robustness and reproducibility through peer review of the computational methods and validation processes are also emphasized, commonly including cross-validation with experimental or observational data.
Mechanistic relevance and predictive capacity are also crucial, often requiring the tool to demonstrate meaningful predictions of endocrine disruption endpoints, including potential integration with in vitro to in vivo extrapolation (IVIVE) frameworks and multi-endpoint test batteries.
The document also highlights the importance of continuous updating and improvement, incorporating new toxicological data as they become available, to keep models current within the evolving chemical space of the EDSP universe. Transparency and documentation are also essential, including a clear rationale for methodology, assumptions, limitations, and performance metrics.
While a detailed EPA EDSP-specific document was not directly found, these general principles follow validation best practices referenced in the context of toxicology tools development and evaluation aligned with EPA data resources and decision frameworks. The emphasis on comprehensive and curated databases, verification of underlying test data, and explicit applicability domains is a cornerstone in EDSP-related tool validation. Additionally, there is recognition of the need for mechanistic understanding and test battery validation, especially for complex endpoints like endocrine disruption.
The guidance in the document pertains to the Endocrine Disruptor Screening Program universe of chemicals and is intended for use in chemical prioritization within the EDSP. However, the document does not specify the exact computational toxicology tools to be used. It serves as a valuable resource for ensuring the reliable and transparent use of computational toxicology tools in the EDSP, contributing to the protection of human health and the environment.
[1] Reference for validation best practices [2] Reference for mechanistic understanding and test battery validation
- The document underscores the relevance of science, particularly in the field of toxicology, as it guides the validation of computational tools for assessing the potential endocrine-disrupting effects of chemicals, which has significant implications for medical-conditions and public health.
- In the quest to ensure the reliability of computational toxicology tools for the Endocrine Disruptor Screening Program, the principles outlined in the document emphasize the importance of advanced technology, such as artificial intelligence and machine learning, to model and predict the behavior of chemicals, thereby informing decision-making processes related to medical-conditions, science, and environmental safety.
%20and%20Guiding%20Validation%20Standards){kind=link}