Conceptual

Deep Learning Methods in Protein Bioinformatics and Protein Design

A survey-level overview of how deep learning is applied across protein bioinformatics, organized into three problem categories: structural prediction (3D structure from amino-acid sequence, e.g. AlphaFold2), functional prediction (properties, interactions, binding sites), and protein design (proposing sequences or structures meeting a functional objective, including de novo design). Covers the sequence-structure-function paradigm, the deep-learning methodologies used in each category, and how prediction advances feed into design.