Data-driven regression analysis of amylose using Sombor molecular descriptors

Area of Science:

• Topological chemistry and molecular modeling.
• Biochemistry and polysaccharide analysis.

Background:

• Amylose, a key starch component, is crucial for plant energy storage and has nutritional implications.
• Topological molecular analysis offers insights into molecular structure and properties.

Purpose of the Study:

• To analyze amylose structure using novel Sombor indices.
• To explore the relationship between topological indices and physicochemical properties.
• To identify the most effective Sombor indices for predicting amylose behavior.

Main Methods:

• Application of Euclidean geometry to derive area-based, perimeter-based, and angular-based Sombor indices.
• Regression analysis using supervised machine learning to correlate indices with properties.
• Analysis of Variance (ANOVA) to determine statistical significance.

Main Results:

• Introduction of the first area-based Sombor index.
• Identification of SO[Formula: see text] and SO[Formula: see text] as potent predictors of amylose properties.
• SO[Formula: see text] demonstrated highest predictive accuracy and model robustness.

Conclusions:

• Sombor indices are reliable topological descriptors for molecular classification.
• This approach enhances understanding of amylose's physicochemical behavior.
• Findings support applications in biopolymer analysis, materials science, and food technology.