A Graph-Theoretical and Machine Learning Approach for Predicting Physicochemical Properties of Anti-Cancer Drugs

Topological graph theory provides a quantitative approach to understanding the structural complexities of sulfonamide compounds, which are prominent for their therapeutic importance in cancer treatment. A new computational scheme to predict the physicochemical and biological functions of sulfonamide derivatives, based on connection numbers and connection-based topological indices as alternatives to the theoretically overt degree-based index, is proposed. A set of structurally diverse sulfonamide compounds as chemical graphs is considered, and the relevant graph descriptors are computed using different connection numbers. Due to the complexity of the calculations involved in connectivity and other such indices, algorithms were developed in Python 3.12.12 to automate the extraction and calculation of these indices. QSPR analysis, with the help of supervised machine learning models like linear regression, among others, and various statistical techniques, was employed to obtain insight into the relationships existing between the structural properties and the molecular properties measured, such as melting point, molecular weight, etc. These results demonstrate the great predictive capability of connection-based indices in assessing pharmacologic efficacy or molecular behavior. The holistic setting thus links topological modeling to data-driven prediction and provides a window into the rational design and optimization of sulfonamide-based cancer therapeutics.