How do I use Multi-Parameter Optimization (MPO) for binary classification in MolSoft's ICM-Pro?

[Andrew Orry]

Q How do I use Multi-Parameter Optimization (MPO) for binary classification in MolSoft's ICM-Pro?

A: To use MPO for binary classification in MolSoft's ICM-Pro, follow the instructions in the documentation here. 

The key steps are:

1. Select a Binary Classification Column – Choose the dataset column that distinguishes between two classes (e.g., active vs. inactive).
2. Choose Molecular Descriptors – Identify relevant physicochemical and molecular properties that contribute to classification.
3. Perform Feature Selection – Use Random Forest classification to determine the most important descriptors.
4. Generate Initial MPO Parameters – Set up the MPO model based on the selected features.
5. Optimize MPO Parameters – Adjust weights and thresholds to maximize classification performance, typically measured by the Area Under the Curve (AUC).