SMILES in QSPR/QSAR Modeling: results and perspectives

Area of Science:

• Computational Chemistry
• Cheminformatics
• Medicinal Chemistry

Background:

• Quantitative Structure-Property/Activity Relationship (QSPR/QSAR) studies are crucial for predicting molecular properties and activities.
• Simplified molecular input line entry system (SMILES) provides a linear notation for chemical structures, enabling computational analysis.
• Developing accurate molecular descriptors is key to enhancing the predictive power of QSPR/QSAR models.

Purpose of the Study:

• To describe a technique for constructing optimal descriptors using SMILES notation.
• To compare the performance of SMILES-based optimal descriptors against those derived from molecular graphs in QSPR/QSAR modeling.
• To present detailed QSPR/QSAR models for various chemical and biological properties.

Main Methods:

• Calculation of optimal molecular descriptors utilizing the Simplified molecular input line entry system (SMILES).
• Comparison of SMILES-based descriptors with descriptors derived from molecular graphs (hydrogen-filled graphs, atomic orbital graphs).
• Development and validation of QSPR/QSAR models for predicting normal boiling points, mutagenicity, toxicity, and anti-HIV-1 potentials.

Main Results:

• Demonstration of a method for generating optimal descriptors from SMILES strings.
• Comparative analysis showing the efficacy of SMILES-based descriptors in QSPR/QSAR modeling.
• Successful QSPR/QSAR models were developed for normal boiling points of organic compounds, mutagenicity of heteroaromatic amines, toxicity, and anti-HIV-1 potentials of TIBO and HEPT derivatives.

Conclusions:

• SMILES-based optimal descriptors offer a viable and effective approach for QSPR/QSAR analyses.
• The developed QSPR/QSAR models highlight the predictive capabilities of SMILES descriptors for diverse properties.
• Further improvements to the SMILES-based concept can enhance future QSPR/QSAR studies.