Proceedings of the International scientific and practical conference “Science in the Modern World” (January 19-21, 2026) / Publisher website: www.naukainfo.com. - Cambridge, United Kingdom, 2026. - 203 p.

7 static and probabilistic, requiring integration with experimental and dynamic methods to fully capture protein function [1, pp. 4-6]. The determination of protein three-dimensional structure is a central challenge in modern biology and biotechnology, as protein structure underpins molecular function, stability, and interactions. Structural alterations, even minor ones, can profoundly affect biological activity and are implicated in numerous diseases. Consequently, accurate structural information is foundational for drug discovery, enzyme engineering, and molecular biology [4, p. 1886]. Traditional experimental techniques - X-ray crystallography, nuclear magnetic resonance (NMR) spectroscopy, and cryo-electron microscopy (cryo-EM) - have enabled major advances but suffer from intrinsic limitations. Crystallography requires successful protein crystallization, often unachievable for membrane or intrinsically disordered proteins; NMR is constrained by protein size and sample requirements; and cryo-EM relies on costly instrumentation and specialized expertise [4, p. 1887]. These constraints make experimental structure determination expensive, slow, and difficult to scale, motivating the development of computational alternatives. Recent advances in artificial intelligence (AI), particularly deep learning, have reshaped the analysis of biological data by enabling models to extract complex patterns from large-scale genomic and proteomic datasets [2, pp. 6272-6274]. In structural biology, neural network-based approaches have demonstrated unprecedented success in predicting protein folds, interactions, and functional features directly from sequence data [2, pp. 6275-6277]. Unlike earlier computational approaches that relied heavily on physics-based simulations or fragment assembly, modern AI models integrate evolutionary, geometric, and statistical information within end-to-end learning frameworks. This shift has positioned AI not as a supplementary tool, but as a core driver of innovation in protein structure prediction [5, pp. 707-708]. Developed by DeepMind, AlphaFold was designed to address the long-standing protein folding problem: predicting a protein’s three-dimensional structure solely from its amino acid sequence [3, pp. 584-585]. Its breakthrough lies in the use of a