Latent Analysis

The latent module reduces high-dimensional protein features into interpretable axes.

Feature construction

Features include:

• EC50, Δmax, Hill, R²
• NSS
• Scaled versions of all parameters
• Optional redox-related features

All features are standardized before analysis.

Methods

Principal Component Analysis (PCA)

• Captures directions of maximum variance
• Useful for visualization and clustering

Factor Analysis (FA)

• Extracts latent factors
• Often more interpretable when features are noisy

Outputs

• Latent coordinates per protein
• Loadings for each feature
• Explained variance (for PCA)

Interpretation

• Latent axes represent underlying biological trends
• Loadings reveal which parameters drive variation
• Clusters in latent space suggest distinct response modes

This step compresses complex behavior into interpretable structure.