Application of mechanistic data in chemical assessment -Presentation
On this page:
Background: Literature based chemical assessments have adopted systematic review methods and systematic maps to ensure rigor, transparency, and identification of all the evidence relevant to answering a question. This is a significant semantic challenge because different communities use different words to describe the same things, making it difficult for people (and computers) to find and understand the characteristics (e.g. outcomes, exposures, methods, test articles, etc.) of an evidence base and relationships in between. In our experience with epigenetic research there are no guidelines to data reporting and author terms (e.g. methylation, DNA methylation, hypermethylation, hypomethylation, histone methylation, differential DNA methylation) are inconsistently used which may lead to duplication and/or misrepresentation of study findings. Objective: The objective of this work is to discuss the linguistics challenge presented by the adoption of systematic reviews and systematic maps as a change in use of evidence supporting a literature based chemical assessment. Using an example from epigenetic research this work will introduce semantics and ontologies in a translational approach to indexing findings across literature resources while managing those ontologies in a flexible framework that can be used as a point of integration across databases of toxicological findings. Discussion: Ontologies can serve as a point of integration across various databases of toxicity findings, and semantic maps have implications for machine learning approaches to identifying relationships between exposure and effect (e.g. adverse outcome pathways/networks) that might have been missed because of our dependence on humans to propose them. This work presents a solution to this challenge using an example from epigenetic data that was semantically translated and organized into a common framework. Semantic matching and ontologies enhance systematic methods for evaluating environmental chemical health risks and permit a data-driven framework supporting evidence synthesis and integration interpretable by both humans and machines.