Application of data mining approach to identify drug subclasses based on solubility and permeability

Area of Science:

• Pharmacology
• Computational Chemistry
• Drug Development

Background:

• Drug intestinal absorption is governed by solubility and permeability, forming the basis of the Biopharmaceutics Classification System (BCS).
• Current BCS boundaries for permeability and solubility are under scientific review, necessitating refined classification methods.

Purpose of the Study:

• To develop a predictive model for human drug permeability using data mining and molecular descriptors.
• To classify drugs into subclasses based on dose/solubility ratio (D/S) and effective permeability (Peff).

Main Methods:

• Data mining analysis of a selected drug dataset.
• Development of a human permeability predictive model using molecular descriptors.
• Clustering and classification analysis based on D/S and Peff.

Main Results:

• Predicted Peff values for 30 drugs showed good agreement with reported fraction of drug absorbed.
• Classification analysis revealed that Peff predominantly influences drug subclasses over D/S.
• Identified two Peff cut-off values (1 × 10⁻⁴ and 2.7 × 10⁻⁴ cm/s), defining an intermediate permeability group.

Conclusions:

• Advanced computational analysis successfully identified complex relationships between physicochemical properties and drug bioperformance.
• The study provides a refined approach to drug classification based on permeability and solubility, aiding in drug development.