Cheminformatics and Machine Learning-Driven QSAR Analysis of SPHK2 Inhibitors for Anticancer Drug Design

Area of Science:

• Computational chemistry and cheminformatics
• Drug discovery and medicinal chemistry

Background:

• Sphingosine kinase 2 (SPHK2) is crucial in sphingolipid metabolism and a target for cancer therapy due to its role in tumor growth and resistance.
• Targeting SPHK2 offers a potential strategy for developing novel anti-cancer agents.

Purpose of the Study:

• To identify and characterize novel SPHK2 inhibitors using a computational approach.
• To develop predictive quantitative structure-activity relationship (QSAR) models for SPHK2 inhibitors.

Main Methods:

• Analysis of 269 SPHK2 inhibitors from ChEMBL using molecular descriptors (PubChem, MACCS, CDK, Substructure, Klekota-Roth).
• Development and validation of QSAR models using six machine learning algorithms (Random Forest, etc.) with metrics like R2 and Q2.
• Evaluation of top-ranked compounds using Glide docking and MMGBSA energy calculations to assess binding affinity.

Main Results:

• Random Forest models with Klekota-Roth and PubChem descriptors achieved the best predictive performance (R2 = 0.78, Q2 = 0.72).
• Lead compounds (CHEMBL2409888, CHEMBL2409889) displayed significant binding affinities (ΔG ≈ -58 kcal/mol) to key residues (GLU293, ARG652, HIS289).
• Inactive compounds exhibited weaker binding interactions, validating the predictive models.

Conclusions:

• An integrative computational pipeline successfully identified potent SPHK2 inhibitors with key molecular features.
• The study provides a rational framework and valuable candidates for further optimization and biological validation of SPHK2 inhibitors in cancer treatment.
• Findings underscore the utility of combining machine learning and molecular modeling for drug discovery.