Simulations Plus Collaborates with Bayer, Roche, and Others to Expand Chemical Space and Applications of pKa Prediction Model

Jul 11, 2024 18:20 CST Updated 18:20

Simulations Plus

Provider of drug research and development, educational software, and related consulting services

Bayer

Pharmaceutical Product R&D Developer

Roche

Oncology Drug Research, Development, and Manufacturing

Win-Win Cooperation

Simulations Plus Collaborates with Bayer, Roche to Expand Chemical Space and Applications for pKa Prediction Model

In May 2024, several scientists from Simulations Plus, Bayer Pharmaceuticals, Roche Pharmaceuticals, Genentech, and Bayer Crop ScienceMolecular InformaticsA paper titled:Best of both worlds: An expansion of the state of the art pKa model with data from three industrial partnersThe article.This article describes that five companies collaborated to develop version 11.0 of the pKa prediction model S+ pKa. Simulations Plus was responsible for the development and optimization of the model, while the other four companies provided their compound structures and corresponding pKa experimental data (some compounds had multiple pKa values). As a result, the prediction accuracy and chemical space of the final model were significantly improved.

Tens of thousands of industrial compounds

S+pKa Version 11.0 (Built into the commercial software ADMET Predictor Version 11.0), is the only product in the industry that adoptsTens of thousands ofThe commercially available pKa prediction model, S+pKa Version 11.0, was developed using real and reliable experimental data of compounds from the industrial sector, and it incorporates a total of 50,514 compounds (including 70,669 pKa experimental values).

Table1：S+pKa Model v7.0 And v11.0 Summary of Performance Improvements Between

Integrated 10 Sub-models

The S+pKa model uses an Artificial Neural Network Ensemble (ANNE) of 10 different types of ionizable atoms to predict microscopic ionization constants.Ensembling multiple models will enhance the prediction accuracy of the integrated model.

1. Hydroxy acids

2. Acidic amides

3. Acids of aromatic NH

4. Thioacids

5. Carboacids

6. Amines

7. Bases of aromatic N

8. N-oxides

9. Thiones

10. Carbobases (protonable C in certain π-excessive rings)

Figure1：S+pKa Version11.0The model does not include protonatable carbon atoms.(Carbobases)Prediction Performance of Submodels

Chemical Space of the Model

Compounds with a molecular weight greater than 600 Da provided by Roche and those with a molecular weight greater than 900 Da provided by Genentech have expanded the chemical space of the model.

Figure2：Used for buildingS+pKa Model Compound Molecular Weight andpKaHistogram of Observed Values

Application of the Model

The well-developed S+pKa model (version 11.0) has been applied in three pharmaceutical companies and one agrochemical company as follows,

Bayer:S+pKa Model Integrated into Bayer's ADMET Prediction Platform
Roche:The S+pKa model is used in various stages of drug discovery, including the early development of formulation prescriptions.
Genentech:The S+pKa model is now used as an input value for many predictive software programs within the company and helps prioritize compounds before proceeding to further testing.
Bayer CropScience:The S+pKa model is used to predict the pKa of new agrochemical compounds, thereby assessing the bioavailability of these compounds in plants and pests.

In addition, the S+pKa model has been integrated/updated into the commercial software ADMET Predictor 11.0.

Conclusion

The best strategy to improve the accuracy and precision of predictive models is to provide the model development team with representative experimental data from within the industry. The more data sources available and the more robust the experimental methods, the better the predictive performance of the model.

Moreover, if the developed predictive models can be made available to the public (commercial or open-source), these models will benefit more people compared to individual models developed internally by companies or academic institutions.

Article Search

This Official Account Has Launched the Function to Search Any Past Articles!

Entering "search" in the menu bar of the WeChat Official Account allows you to search past WeChat content.