Analysis of A Drug Target-based Classification System using Molecular Descriptors

Area of Science:

• Computational chemistry
• Pharmacology
• Bioinformatics

Background:

• Drug-target interaction is crucial for drug discovery and repositioning.
• The KEGG database classifies drugs based on their targets.
• Understanding these interactions aids in developing new therapeutics.

Purpose of the Study:

• To investigate drug classification across five target-based classes: G protein-coupled receptors, Nuclear receptors, Ion channels, Enzymes, and Pathogens.
• To identify key molecular descriptors that define these drug-target relationships.
• To develop an accurate predictive model for drug target classification.

Main Methods:

• Utilized molecular descriptors to represent drug compounds.
• Employed Maximum Relevance Minimum Redundancy (mRMR) and Incremental Feature Selection (IFS) for feature selection.
• Constructed a predictive model using the Nearest Neighbor algorithm.

Main Results:

• Successfully identified significant molecular descriptors for drug target classification.
• The Nearest Neighbor model achieved optimal performance in distinguishing between the five target-based classes.
• Key descriptors were highlighted for their importance in predicting drug-target interactions.

Conclusions:

• Molecular descriptors can effectively classify drugs into target-based categories.
• The developed model provides a robust method for predicting drug-target classes.
• This research contributes to advancing drug discovery and repositioning strategies.