Ligand-Based Approach for Multi-Target Drug Discovery: PTML Modeling of Triple-Target Inhibitors

Area of Science:

• Computational chemistry and drug discovery
• Oncology research
• Machine learning in pharmacology

Background:

• Cancer is a complex multi-genetic disease requiring multi-target drug discovery approaches.
• Computational methods accelerate the identification of multi-target anticancer agents.
• This study focuses on designing inhibitors for Tropomyosin Receptor Kinase A (TRKA), poly[ADP-ribose] polymerase 1 (PARP-1), and Insulin-like Growth Factor 1 Receptor (IGF1R).

Purpose of the Study:

• To rationally design and predict triple-target inhibitors against TRKA, PARP-1, and IGF1R using computational methods.
• To combine a perturbation-theory machine learning model (PTML-EL-MLP) with Fragment-Based Topological Design (FBTD).
• To generate new chemical diversity for multi-target drug discovery in oncology.

Main Methods:

• Utilized ChEMBL database for chemical and biological data extraction.
• Applied the Box-Jenkins approach to generate multi-label topological indices.
• Developed a perturbation-theory machine learning model ensemble of multilayer perceptron networks (PTML-EL-MLP).
• Integrated PTML-EL-MLP with Fragment-Based Topological Design (FBTD) for molecular design.

Main Results:

• The PTML-EL-MLP model achieved approximately 80% accuracy.
• FBTD enabled physicochemical and structural interpretation of the model.
• Identified key molecular fragments positively influencing multi-target activity.
• Virtually designed four new drug-like molecules predicted as triple-target inhibitors.

Conclusions:

• The combination of PTML modeling and FBTD is effective for generating novel chemical entities.
• This approach facilitates multi-target drug discovery in oncology.
• The study highlights the potential for creating new chemical diversity in drug development.