Deep Neural Network–Based Identification of Protein Secondary Structures and Structural Complexities

Le résumé fourni par la source

Proteins are essential biological macromolecules responsible for a wide range of cellular functions, including molecular transport, enzymatic catalysis, and signal regulation. Understanding protein structure is crucial for analyzing protein behavior and supporting drug discovery and biomedical research. Protein structure prediction (PSP) aims to determine the three-dimensional conformation of a protein based solely on its amino acid sequence. Despite the presence of only twenty standard amino acids, an enormous diversity of protein structures arises from their varied arrangements and interactions. Protein conformations are highly sensitive to external influences such as chemical compounds, drugs, and environmental conditions, which can induce structural changes affecting biological functionality. Accurate prediction of protein structures therefore remains a challenging computational problem. Protein secondary structures mainly consist of α-helices, β-sheets, and loop regions, and variations in their proportions introduce additional complexity during prediction. In this work, a deep neural network–based framework is proposed for identifying protein secondary structures and detecting structural complexities that arise during prediction. The proposed approach focuses on improving prediction reliability by addressing redundancy and structural ambiguity. The methodology is evaluated using a large protein dataset and demonstrates its potential to support researchers in drug development and biomedical applications by providing accurate and scalable protein structure analysis.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.