(429d) QSAR Study of Combretastatin-like Chalcones As Cancer Cell Growth Inhibitors Using Linear and Non-Linear Machine Learning Approaches

Combretastatin-like chalcones are known to be effective inhibitors for the mitotic process that tends to help the growth of cancer cells. Chalcones interact with 𝝱-tubulins, and inhibit the polymerization process that ends up increasing tumor size [1]. Based on the works of Ducki et al [2], 87 chalcones were selected. It is important to mention that an attempt to model inhibitory concentration (pIC₅₀) was made by Lipinski et al [3]. Although they have achieved models with acceptable fitness, there is still room for improvements. It can be noticed in their work that, although they used a variety of descriptor classes, their use of model development algorithms is lacking in light of recent advances.

This project aims at modifying, improving, and implementing machine learning and model development algorithms such as Artificial Bee Colony, Multi-gene Genetic Algorithm, Particle Swarm Optimization algorithms in combination with linear, non-linear, and LASSO regression algorithms. The goal is to develop models with better fitness using descriptors that are comparatively easier to calculate (e.g. connectivity descriptors, 2D descriptors). Avogadro open access software was used to develop and optimize the molecular structures, and Dragon 6.0 software was used to generate descriptors for the given compounds. One-fifth of the samples were used for external validation of model. Model superiority was decided based on values of R², Q², MSE, and MAE values of the models.

Selected References:

[1] Sharma et al, A review of mechanisms of antitumor activity of chalcones, 2016, Anticancer Agents Med Chem, 16, 200-211.

[2] Ducki et al, Combretastatin-like chalcoles as inhibitors of microtubule polymerization. Part 1: synthesis and biological evaluation of anitvascular activity, 2009, Bioorg Med Chem, 17, 7698-7710.

[3] Lippinski et al, A molecular modeling study of Combretastatin-like chalcones as anticancer agents using PLS, ANN and consensus models, 2018, Struct Chem (2018), https://doi.org/10.1007/s11224-017-1072-2