PROTEIN THREADING

Protein threading also referred to as Fold Recognition is used in the prediction of protein structures from aminoacid sequence.
Target sequence (the protein structure for which the stucture is being predicted) is threaded into the backbone structures of a collection of template protiens. The collection of template proteins is referred to as Fold Library.
Then a 'Goodness of Fit' score is calculated for each structure sequence alignment. Goodness of Fit is often derived in terms of an empirical energy function, based on statistics derived from known protein structures . Many other scoring function have been proposed and the most useful among them are Pairwise terms (interaction between pairs of aminoacids) and solvation terms.
Threading methods share some of the characteristics of both homology modelling and ab initio prediction methods.

Fold recognition methods can be broadly divided into 2 types:

• Methods that derive a one-dimensional (I-D) profile for each of the protein structure in the fold library and align target sequence to these profiles.
• Methods that consider the full three dimensional structure of the protein template.

A simple way of profile representation would be to take each aminoacid in the structure and simply label it according to whether it is buried in the core structure of the protein or exposed on
the surface. More eloborate profiles might take into account the local secondary structure (ex: whether the aminoacid is part of an alpha helix) and/or evolutionary information (ex: how conserved the amino acid is).
In the three dimensional representation, the structure is modelled as a set of interatomic distances i.e. distances are calculated betweem some or all the atom pairs in the structure. This is a much better and flexible description of the structure, but is is much harder to use in calculating an alignment.
The profile based fold recognition aproach was first described by Bowie, Luthy and Eissenberg in 1991.
The term Threading was first coined by Jones, Tailor and Thornton in 1992, and originally referred specifically to the use of a full three dimensional structure atomic representation of the protein template in fold recognition. Today the two terms are frequently used interchangeably.
Fold recognition methods are more frequently used effective because it is believed that there are a strictly limited number of different protein folds in nature, mostly as a result of evolution and also due to constraints imposed by the physics and chemistry of polypeptide chains. there is, therefore a good chance that a protein which has a similar fold to the target protein has already been studied by X-ray cristallography/NMR spectroscopy and can be found in the ProteinDataBank (PDB). Currently there are just about 1100 different protein folds known and , but new follds are still being discovered every year.
Many different algorithms have been proposed for finding the correct threading of a sequence onto a structure though many make use of Dynamic Programming in some form. For full threedimensional threading, the problem of identifiying the best alignment is very difficult and researchers have made use of many combinatorial optimization methods such as simulated annealing or branch and bound searching to arrive at heuristic solutions.

RAPTOR is an integer programming based protein threading software