See pptx for visual description and Jensen et al., 2023 for details:
Jensen, M. A., Blatz, D. J., & LaLone, C. A. (2023). Defining the biologically plausible taxonomic domain of applicability of an adverse outcome pathway: A case study linking nicotinic acetylcholine receptor activation to colony death. Environmental Toxicology and Chemistry, 42(1), 71-87.
Domain of applicability for Key Events (KE) in AOP-Wiki Tables:
List species, sex, life stage from empirical studies (in vitro and in vivo)
Define list of species for biologically plausible taxonomic domain of applicability (in silico) - can be pulled from 3rd party tools (e.g., SeqAPASS)
Domain of applicability for Key Event Relationships (KER) in AOP-Wiki Tables with evidence in free text:
Define Plausible Taxonomic Domain of Applicability where KEup list of species matches KEdown list of species (narrowest DOA; Jensen et al., 2023)
-This info is derived from empirical studies that support the KEs and through lines of evidence toward structural and functional conservation for plausible taxonomic DOA
-Life stage and Sex can be combined with taxonomic tDOA
-Machine readable tables with links out to details for evidence calls in free text
-Ensure DOA categories are linked
To help the AOP-framework to better accommodate and apply to different stressor types (e.g., chemicals, radiation, viruses, immunotherapies...), consider adding additional types of stressor tables that would allow descriptors of non-chemical stressor types to be added.
The SAAOP is hosting an online workshop series to obtain perspectives and input from a broad range of AOP community stakeholders in an effort to define the features and functions (requirements) to be implemented as part of AOP-Wiki 3.0, with as much specificity as possible.
Please “pitch” your desired features here prior to the sessions. Vote for the ideas to be discussed during the workshop sessions using the star and discussion features in the forum.
Session 1: July 26, 2023
• 9:00-12:00 (GMT-4:00) North America and Europe [Register here!]
• 20:00-22:00 (GMT-4:00) North America and Asia [Register here!]
• Session 2: August 16, 2023
• 8:00-10:00 (GMT +2:00) Europe and Asia [Register here!]
• 15:00-18:00 (GMT +2:00) Europe and North America [Register here!]
here are a couple of points discussed in today's webinar with regards to term standardization, ontologies and visualization for making the AOP-Wiki more intuitive.
Issues
People do not fill in important information in tables. Instead, they put information in free text
People use generic terms
Although the process is driven by ontologies, sometimes inappropriate terms are inserted
Duplicated concepts exist
Data ossification: who maintains older AOPs, KEs when a new functionality is added
Ontology proliferation or custom-made vocabularies
Missing terms from major Ontologies
Standardization of KE names
Stressors may be linked to more than one chemical (eg stressor 57).
Stressors may not be Chemicals, e.g SARS-COV-2 . Also allow for other types of stressors to be added, e.g radiation
Increase Interoperability with regulator world
Here is a list of suggestions in order to tackle some of these issues above.
Low hanging fruit
AO names could be mapped to MedDRA
Standardize and fix terms in categories/Stressors
Standardize and fix terms in categories/Organisms
OLS-like interface for ontology consolidation
Visualization of AOP networks will help identify duplication and hubs (already implemented for 3rd party tool page, but ideally needs to be in every AOP/KE page)
Mid-term
Liaise with ontologies that are open to term suggestions, e.g UBERON
Treat the project in an open-source way/Bug tracking for missing or incomplete information
Treat the project in an open-source way/CI Reports for automatically identified common problems
Treat the project in an open-source way/Governance: Create a body that may authorize proposed changes in case the AOP author is unresponsive
Requiring policy changes
Standardize and fix terms in categories; Stressors must have only a PubChem ID and not Name, Preferred Name, User Term
Put Biological Organization, Cell, Organ in KE component table (allows for multiple rows)
Allow for Tissue and Subcellular location (relevant for MIE)
Since 2018 we have been monitoring the growth of the AOP-Wiki in terms of user defined AOPs, relationships, events, emergent AOPs, etc. This information is reported out at least biannually, but requires an expert user to extract this information from the XML using custom R code. Would be nice to build in functions that calculate these statistics automatically and can provide real-time read out and cumulative plots, etc. on an on-going basis.
AOPs are envisioned as a way to organize information not just relevant to chemical safety assessment, but also relative to radiation, disease/viruses, even economic stress. As more of these examples get populated into the wiki, it may be useful to have ways to sort/filter based on general application domain areas.
We need a solution for the review of AOP that have shared KEs and KERs.
This is especially important if these KEs/KERs have been borrowed from AOPs that have already been reviewed.
One suggestion is that during the review, each KE and KER gets reviewed independently of each other, and each have their own "review documentation" attached to them.
This way, any review conducted will "follow" that KE/KER wherever they go. This is the same as all the other information related to KEs and KERs
i.e. We should consider making AOP reviews "modular" as well. Or rather, a "review" is an attribute/property for each KE, KER and AOP in KB.
just like AOPs are make up of KEs and KERs, the "AOP Review" can be "made of" its component "KE Reviews" and "KER reviews", along with any higher level synthesis that occurs at the AOP level.
All of these "sub-reviews" can be conducted by the same group of reviewers (as is currently done) or can optionally be conducted individually, as needed.
I've included an image describing this idea (hope it makes sense :))
A suggestion to tag AOPs that have sufficient quantitative understanding or aligned computational models to facilitate their use for quantitative applications (suggestion from webinar 1).
