Extracting prime protein targets as possible drug candidates: machine learning evaluation

Area of Science:

• Computational Chemistry
• Drug Discovery
• Bioinformatics

Background:

• Identifying potent drug targets is crucial for developing new MRSA therapies.
• Molecular docking and machine learning are key tools in drug discovery.

Purpose of the Study:

• To identify prime protein targets (PPTs) for MRSA drug candidates.
• To develop an efficient algorithm for extracting high-affinity ligands (HALs) and PPTs.

Main Methods:

• Combined data from 10 molecular docking evaluations.
• Utilized machine learning for data mining and modeling (DDM) and reverse modeling (RM).
• Employed K-means clustering (KMC) and validated against Gaussian mixture model (GMM) and DBSCAN.

Main Results:

• Identified 5 high-fidelity PPTs (17%) that bind with 19% of ligands.
• PPT2 (41.1% average HPC), PPT14 (25.46%), and PPT15 (23.12%) are top candidates.
• KMC and GMM proved effective for clustering, while DBSCAN struggled with noise.

Conclusions:

• The developed algorithm efficiently identifies MRSA drug candidates.
• This approach is applicable to various pathogens and effective with sparse data.
• Predicts promising next-generation MRSA drug candidates.