TOPS-MODE versus DRAGON descriptors to predict permeability coefficients through low-density polyethylene

Area of Science:

• Computational chemistry and cheminformatics.
• Materials science, focusing on polymer permeability.
• Quantitative Structure-Property Relationships (QSPR).

Background:

• Understanding compound permeability through polymers is crucial for material selection and design.
• Existing QSPR methods often struggle to achieve high predictive accuracy and interpretability.
• The low-density polyethylene (LDPE) system serves as a relevant model for barrier properties.

Purpose of the Study:

• To apply the TOPological Sub-Structural MOlecular DEsign (TOPS-MODE) approach to model permeability coefficients.
• To compare the predictive power of TOPS-MODE against eight other descriptor-based methods.
• To assess the structural interpretability of the developed TOPS-MODE model.

Main Methods:

• Utilizing the TOPS-MODE methodology for molecular descriptor generation and model building.
• Experimental determination or collection of permeability coefficients for 38 organic compounds through LDPE at 0°C.
• Comparative analysis against models built using constitutional, topological, BCUT, 2D autocorrelations, geometrical, RDF, 3D Morse, and GETAWAY descriptors.

Main Results:

• A TOPS-MODE based model explained over 92% of the variance in experimental permeability coefficients.
• Alternative methods, using the same number of descriptors, explained less than 75% of the variance.
• The TOPS-MODE approach identified specific molecular fragments contributing to permeability.

Conclusions:

• TOPS-MODE significantly outperforms traditional descriptor methods for predicting LDPE permeability.
• The approach provides enhanced structural interpretability, aiding in the understanding of permeability mechanisms.
• TOPS-MODE offers a powerful tool for designing molecules with tailored permeability properties.