https://orcid.org/0000-0001-9712-4621
Abstract
Background:
To prioritize compounds with a higher likelihood of success, artificial intelligence models can be used to predict absorption, distribution, metabolism, excretion and toxicity (ADMET) properties of molecules quickly and efficiently.
Methods:
Models were trained with BioPrint database proprietary data along with public datasets to predict various ADMET end points for the SAFIRE platform.
Results:
SAFIRE models performed at or above 75% accuracy and 0.4 Matthew’s correlation coefficient with validation sets. Training with both proprietary and public data improved model performance and expanded the chemical space on which the models were trained. The platform features scoring functionality to guide user decision-making.
Conclusion:
High-quality datasets along with chemical space considerations yielded ADMET models performing favorably with utility in the drug discovery process.
Tweetable abstract
BioPrint meets the artificial intelligence age: researchers trained absorption, distribution, metabolism, excretion and toxicity machine learning models with the BioPrint database for the new SAFIRE platform.
Author contributions
CR Sage and LM Goncalves were responsible for the conception and design of the work. SE Biehn and CR Sage wrote the manuscript, and SE Biehn generated all figures for the manuscript. SE Biehn, S Ramirez and C Mueller were responsible for dataset curation and retrieval and model development. SE Biehn created all models for the SAFIRE platform in Python. J Lehmann provided analysis and conceptual research along with dataset curation and advised on the analysis and interpretation of data for the work. SE Biehn and J Lehmann assessed publicly available models. JD Marty provided project management structure and advised on analysis and interpretation of data for the work. C Mueller created web service for platform model usage. F Tillier retrieved proprietary data and consulted on the usage, interpretation and analysis of proprietary data. All authors approved the draft.
Acknowledgments
The authors would like to thank all members of the Eurofins DiscoveryAI group for their feedback and insights along with various members of Eurofins Discovery for their feedback and guidance on the platform and manuscript.
Financial disclosure
The authors declare the following financial interests/relationships which might be viewed as potential competing interests: all authors are employees of Eurofins Panlabs, Inc.
Competing interests disclosure
The authors have no competing interests or relevant affiliations with any organization or entity with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending or royalties.
Writing disclosure
No writing assistance was utilized in the production of this manuscript.