Protein sequence-structure threading methods

Jones, D. T., et al., Successful recognition of protein folds using threading methods biased by sequence similarity and predicted secondary structure. Proteins, 1999. 37(S3) p. 104-111. 

This section briefly reviews prediction of the native structure of a protein from its sequence of amino acid residues alone. These methods can be contrasted to the threading methods for fold assignment [Section II.A] [39-47,147], which detect remote relationships between sequences and folds of known structure, and to comparative modeling methods discussed in this review, which build a complete all-atom 3D model based on a related known structure. The methods for ab initio prediction include those that focus on the broad physical principles of the folding process [148-152] and the methods that focus on predicting the actual native structures of specific proteins [44,153,154,240]. The former frequently rely on extremely simplified generic models of proteins, generally do not aim to predict native structures of specific proteins, and are not reviewed here. 

Structure prediction methods have to be validated using comprehensive and representative benchmark sets. A couple of benchmark sets have been proposed [188, 199, 270] consisting of a fold library and a list of structurally similar protein pairs without significant sequence similarity. Such a benchmark [335] involves using the method in question to thread one protein of each pair against the library and to count the number of successful recognitions of the structural match or the number of correct identifications of related protein pairs. A couple of methods have been evaluated this way. 

Threading allows, given a query sequence, to search for sequences with a similar fold but without apparent sequence similarity. Therefore, threading has three major objectives first, provide orthogonal evidence of possible homology for distantly related protein sequences, second, detect possible homology in cases where sequence methods fail, and third, to improve structural models for the query sequence via structurally more accurate alignments. 

Jones, D. T., and C. Hadley. 2000. Threading methods for protein structure prediction. In D. Higgins, and W. R. Taylor (eds.) Bioinformatics Sequence, Structure and Databanks. Heidelberg, Germany Springer-Verlag. 

Protein structure prediction [253,254] is one of the major goals of computational molecular biology. Up to now, homology based and threading methods have been the most successful [185,226,255-264]. However, due to the increasing ratio of the number of known protein sequences to the number of solved protein structures, the development of de novo (or related) methods would be extremely valuable. To date, only limited, nevertheless encouraging, progress has been achieved in such direct approaches [42,254]. Purely de novo predictions are now possible only for peptides... 

On the other hand, there are methods that make use of 3D information instead to find structural templates for sequences. There are many ways to use 3D information to aid prediction. Threading methods exploit the difference in the distribution of inter-residue distances in protein... 

More distantly related structures can be modeled by the use of the process of threading. This is a computational technique by which the sequence of the protein of an unknown structure is threaded through a nom-edun-dant library of well-determined structures to determine which are compatible with the sequence. Basic statistical methods for predicting secondary structure, a helix and p sheet, have 80% veracity. 

---