Version 2024 of Percepta delivers substantial expansions to the training sets and algorithms, improving prediction accuracy for pKa and logP calculators, and a range of ADME and toxicity endpoints. This update also increases coverage of chemical space occupied by new therapeutic modalities such as proteolysis targeting chimeras (PROTACs).
Bracknell, UK (September 30, 2024)—ACD/Labs, an informatics company that develops and commercializes software in support of digitalized R&D, today announced the release of version 2024 Percepta Platform property calculator software. The platform offers molecular property prediction of physicochemical and ADME properties, and toxicity endpoints. These values are used by research chemists to understand in vivo and environmental behaviors of compounds, for the selection of solvent and stationary phase in chromatographic method development, for regulatory submissions in lieu of experimental data; and by computational chemists to build and refine predictive models. Users of Percepta v2024 will benefit from substantial improvements in prediction accuracy for logP, logD, and pKa, and a range of ADME and toxicity endpoints, in addition to increased coverage of novel chemical space.
Users of Percepta v2024 will benefit from substantial improvements in prediction accuracy for logP, logD, and pKa, and a range of ADME and toxicity endpoints, in addition to increased coverage of novel chemical space.
In version 2024 of Percepta, predictive algorithms for acid dissociation constants (pKa), partition coefficients (logP), and distribution coefficients (logD) have been dramatically enhanced and further trained with experimental values from several diverse chemical datasets. The training set expansion included over 2200 experimental pKa values from a pharmaceutical partner, and over 3000 pKa values from other sources, representing a 13% increase of ACD/Labs’ pKa training data set. The logP algorithm has undergone an unprecedented training set expansion of more than 25%, with high-quality data containing novel scaffolds and challenging functional groups. The resulting algorithm improvements enhance prediction accuracy for well-known chemical space, and increase the coverage of complex compounds of pharmaceutical interest, including novel therapeutic modalities such as PROTACs.
“The ability to accurately predict the physicochemical properties of multispecific drugs, such as PROTACs, could be pivotal in ensuring they are successfully optimized for oral administration,” said Andrius Sazonovas, Director of Percepta Solutions Product Development. “To illustrate the ability of the latest version of Percepta to accurately calculate pKa values of these compounds we tested a total of 253 PROTACs and their precursors (which had 491 experimental pKa values). From v2023 to v2024 we see a significant reduction in average prediction errors—from 0.52 to 0.28 log units. Our continued efforts to refine the software mean scientists can be confident in using our tools to support their research.”
“The ability to accurately predict the physicochemical properties of multispecific drugs, such as PROTACs, could be pivotal in ensuring they are successfully optimized for oral administration.”
– Andrius Sazonovas, Director of Percepta Solutions Product Development
Better accuracy in logP, logD, and pKa predictions directly impacts calculators for absorption, distribution, metabolism, excretion, and toxicity that rely on these descriptors (e.g., passive absorption, oral bioavailability, and blood-brain barrier penetration). Additionally, this release delivers up to a 200% increase in the internal training set for various ADME and Tox training libraries—P-gp Substrates, Protein Binding, Ames Test for Genotoxicity, and Aquatic Toxicity. Version 2024 of Percepta also provides a new QSAR model for quantitative prediction of P-gp Efflux, and updates that ensure regulatory filings for observed impurities are aligned with the ICH M7 (R2) Guideline.
“Working with our longstanding customers within top pharma organizations and public data sources, we have carefully curated our modeling space and added new experimental data, to ensure that representative modern chemistry is well covered, for accurate property prediction. We are proud to have our in-silico calculators used at over 80% of top brand pharmaceutical companies, and numerous other organizations, worldwide,” said Daria Thorp, President of ACD/Labs. “Our customers trust ACD/Labs Percepta to choose, register, and optimize lead candidates, meet regulatory compliance requests, ensure robust separation methods, and engage in AI/ML-driven drug design. We are continuously striving to ensure that the best quality and most up-to-date chemistry coverage are available to our users.”
To learn more about v2024 software updates for Percepta, visit www.acdlabs.com/WhatsNew#Percepta
About ACD/Labs
ACD/Labs is a leading provider of scientific software, aiding decision-making and digital transformation in chemical and pharmaceutical R&D.
Our science-centric solutions offer harmonized, AI-ready analytical, structural, and molecular information, with FAIR-compliance. Our cloud-enabled technologies enable the proliferation of chemically intelligent knowledge and workflow automation.
For 30 years, we have partnered with global organizations to enhance productivity and accelerate scientific innovation. For more information, please visit www.acdlabs.com. Follow us on X (Twitter) and LinkedIn.