Protein structure prediction

Protein structure prediction is the computational inference of a protein’s three-dimensional structure from its amino acid sequence and related information. It estimates the arrangement of atoms or residues that underlies the protein’s shape and function.

Protein language model: A machine-learning model trained on protein sequences to capture patterns associated with protein structure and function. Sequence representations from these models can supply information to structure predictors.
Anfinsen's dogma: The principle that a protein’s amino acid sequence determines its native structure under suitable conditions. It frames why structure prediction from sequence is a meaningful scientific goal.
Structural bioinformatics: The use of computational methods to analyze, compare, and model biological macromolecular structures. Structure prediction is one of its central computational tasks.
Homology modeling: A method that builds a protein model using the known structure of a homologous protein as a template. It is especially effective when a close structural relative is available.
Intrinsic disorder: The lack of a stable, independently folded three-dimensional structure in a protein region or protein. Disordered regions challenge predictors built around one well-defined structure.
Multiple sequence alignment: An arrangement of related biological sequences that makes corresponding residues occupy the same columns. Correlated substitutions across aligned homologs can reveal spatial contacts between residues.
Protein Data Bank: An open archive of experimentally determined three-dimensional structures of biological macromolecules. Its accumulated structures supplied training examples and evaluation targets.
Drug discovery: The process of identifying and developing compounds that can prevent or treat disease. Predicted target structures can guide binding-site analysis and early compound design.
Threading (protein structure prediction): A method that fits a protein sequence onto known structural folds to find compatible templates. It can identify plausible folds even when sequence similarity to a template is weak.
Protein conformational ensemble: The collection of interconverting conformations that a protein occupies under specified conditions. A single predicted model may miss functionally important structural variation.

Linked from 17 pages