Prediction of binding affinity for estrogen receptor alpha modulators using statistical learning approaches.

Area of Science:

• Computational Chemistry
• Medicinal Chemistry
• Pharmacology

Background:

• Estrogen receptor (ER) is a key drug target for breast cancer therapy.
• Developing potent and selective ER modulators is crucial for effective treatment.

Purpose of the Study:

• To develop and validate in silico models for predicting ER modulator potency.
• To investigate ligand-receptor interactions for rational drug design.
• To identify key molecular descriptors influencing ER modulator activity.

Main Methods:

• Bayesian-regularized neural network combined with principal component analysis.
• Multiple linear regression (MLR) analysis.
• Quantitative Structure-Activity Relationship (QSAR) modeling using 127 ERalpha modulators.

Main Results:

• High correlation coefficients achieved for the neural network model (training: 0.91, cross-validation: 0.87, test: 0.90).
• MLR model demonstrated reasonable predictivity (R=0.72, Q=0.79) using four molecular descriptors (Xvch6, nelem, SsssCH, SaaN).
• In silico models proved reliable for predicting pIC50 values.

Conclusions:

• The developed in silico models are reliable for predicting the potency of ER modulators.
• These models can facilitate the screening of new compounds and guide the rational design of novel ERalpha modulators.
• The findings contribute to the development of improved breast cancer therapies targeting the estrogen receptor.