Quantitative Structure-Property Relationship (QSPR) Modeling of Drug-Loaded Polymeric Micelles via Genetic Function Approximation

Abstract

Self-assembled nano-micelles of amphiphilic polymers represent a novel anticancer drug delivery system. However, their full clinical utilization remains challenging because the quantitative structure-property relationship (QSPR) between the polymer structure and the efficacy of micelles as a drug carrier is poorly understood. Here, we developed a series of QSPR models to account for the drug loading capacity of polymeric micelles using the genetic function approximation (GFA) algorithm. These models were further evaluated by internal and external validation and a Y-randomization test in terms of stability and generalization, yielding an optimization model that is applicable to an expanded materials regime. As confirmed by experimental data, the relationship between microstructure and drug loading capacity can be well-simulated, suggesting that our models are readily applicable to the quantitative evaluation of the drug-loading capacity of polymeric micelles. Our work may offer a pathway to the design of formulation experiments.

Predicting the Loading Capability of mPEG‐PDLLA to Hydrophobic Drugs Using Solubility Parameters†

“…For example, Wu et al . reported a series of quantitative structure‐property relationships (QSPR) models to evaluate the content of doxorubicin in polymer micelles, but since these models are only for doxorubicin development, it is difficult to generalize to other drugs . In addition, screening the drug carrier by computational simulation requires professional computational software and computational simulation researchers, which is often very expensive.…”

Design of potential anti-melanoma agents against SK-MEL-5 cell line using QSAR modeling and molecular docking methods

“…A good and acceptable QSAR model must obey the following criteria: regression-coefficient (R 2 ) and adjusted regression-coefficient (R 2 adj) values close to one. The Cross validated regression-coefficient (Q 2 cv ) > 0.5, R 2 − Q 2 cv ≤ 0.3, R 2 pred ≥ 0.6, and N test ≥ 5 [33,37,38]. The generated QSAR model satisfied the criteria and therefore acceptable statistically.…”

Insilico Search for Molecules with Enhanced Histone Deacetylase Inhibitory Properties as Templates for Novel Anticancer Agents

“…GFA algorithm, selecting the basic functions genetically, developed better models than those made using stepwise regression methods. And then, the models were estimated using the "lack of fit" (LOF), which was measured using a slight variation of the original Friedman formula, so that best model received the best fitness score [10].…”