Detect predict the fold of a sequence using threading

ebisu group

chemogenomix © 2012

Read more below


The term "threading" (originally the full term was "optimal sequence threading" but this was quickly shortened to just "threading") was first coined in 1991 by David Jones, working with Janet Thornton and Willie Taylor (see Jones, Taylor, Thornton ref. below). Threading is an approach to fold recognition which used a detailed 3-D representation of protein structure. The idea was to physically "thread" a sequence of amino acid side chains onto a backbone structure (a fold) and to evaluate this proposed 3-D structure using a set of pair potentials and (importantly) a separate solvation potential. The program which implemented this method was called THREADER and has been downloaded and used by thousands of users since its first public release in 1994, just after the first Critical Assessment in Structure Prediction meeting was held in Asilomar. At the first CASP, THREADER was the most successful method for fold recognition evaluated at the time, when it correctly identified 8 out of 11 of the folds. THREADER 3 is the latest incarnation of this program which implements many new features (see documentation for details).


What can it do?

Threading is one of the very few methods available which can predict the fold for a protein in the absence of an evolutionary relationship, though if there is an evolutionary link between the target protein and a protein of known structure then there are better methods available e.g. GenTHREADER. Of course if a fold similar to the native fold of the target protein is not in the library, then no fold recognition approach can succeed. Fortunately, certain folds crop up time and time again, and so fold recognition methods for predicting protein structure can be very effective. Unfortunately, the assumption of a rigid backbone structure means that the alignment of the target sequence onto the template can be rather inaccurate. The goods news is that the alignment is generally more accurate for the core regions of the fold, and so with some caution, threading alignments can be of use in providing a seed for a more accurate manual alignment.

Access Manual at UCL site


Threader3 is the latest version in a series of developments originating from a ground breaking paper published in Nature in 1992 that sought to predict the fold of a protein under the assumption that the structural answer was contained in a database of known structures.

Selected References

 Jones, D.T., Taylor, W.R. & Thornton, J.M. (1992) A new approach to protein fold recognition. Nature. 358, 86-89.

Jones, D.T., Miller, R.T. & Thornton, J.M. (1995) Successful protein fold recognition by optimal sequence threading validated by rigorous blind testing. Proteins. 23, 387-397.

Jones, D.T. (1998) THREADER : Protein Sequence Threading by Double Dynamic Programming. (in) Computational Methods in Molecular Biology. Steven Salzberg, David Searls, and Simon Kasif, Eds. Elsevier Science. Chapter 13.

The information described here is taken from the Bioinformatics group web site run by Prof David Jones at University College London