As Candice described in her recent blog “So many pieces of information!” – making actionable decisions is extremely complicated, especially when integrating the results from in silico models alongside historical in vivo studies and in vitro experiments.

When is there sufficient information to make a decision?

Well, it depends the type of decision and the associated risk of making an incorrect prediction.

  • Do I have enough information to support filling data gaps or performing prioritization?
  • Do I have sufficient information to include in a regulatory submission?
  • When should I generate additional experimental data?

Ultimately, this can be answered through an evaluation of the overall confidence in the final assessment. Lower confidence assessments may be sufficient for filling data gaps or the prioritization of chemicals, whereas regulatory submission will require a high degree of confidence in the outcome.

Frameworks that include rules and principles for deriving such assessments of confidence support both a transparent and reproducible evaluation and will ultimately lead to increased acceptance of these methods. One such framework is detailed in the in silico toxicology protocol1 which considers the reliability, relevance and completeness of the information for a set of defined effects/mechanisms.

The framework describes a series of criteria to define what in silico models to include in the first place. Experimental data and/or in silico results may be used to assess these defined effects/mechanisms. A harmonized scoring system (referred to as the Reliability Score) is used to quantify the reliability of the information (based on both experimental data and in silico results) for each defined effect/mechanism.

The framework then outlines the rules and principles for combining this information on the effects and mechanisms based on their defined reliability, their relevance to the final assessment and the completeness of the information.

Development of a consensus for defining these principles is really hard, yet essential to increase the acceptance of these alternative methods across different applications.

The in silico toxicology protocol publication was a good first step; however, we are continually working on these issues through working groups and the publication of case studies.

If you are interested in collaborating on these topics, please reach out to me at gmyatt@leadscope.com or Candice at cjohnson@leadscope.com.

1. Myatt, G.J., Ahlberg, E., Akahori, Y., et al. (2018) In Silico Toxicology Protocols. Regul. Toxicol. Pharmacol. 98, 1-17. doi:10.1016/j.yrtph.2018.04.014

Open access: https://doi.org/10.1016/j.yrtph.2018.04.014

Published by Glenn Myatt

Glenn J. Myatt is the co-founder and currently head of Leadscope (An Instem company) with over 25 years’ experience in computational chemistry/toxicology. He holds a Bachelor of Science degree in Computing from Oxford Brookes University, a Master of Science degree in Artificial Intelligence from Heriot-Watt University and a Ph.D. in Chemoinformatics from the University of Leeds. He has published 27 papers, 6 book chapters and three books.