A method of predicting a protein structure
As the genome sequence databases grow larger and larger, methods for computationally determining a protein's structure from its sequence are becoming very important (the structural genomics problem). It is a protein's structure which determines its function. A threading algorithm is a method used to fit a sequence of an unknown protein to the backbone of a known protein by maximizing the number of good interactions. The following is an overall summary of a threading protocol:
Using homology modelling and secondary structure prediction algorithms, come up with a protein fold which the protein will probably belong to. The fold is selected from a library of backbones. Several backbones are often tested to see which one fits the sequence best.
Then the amino acid sidechains are mapped from the sequence onto the template backbone. This is the "threading" step, where one is essentially threading the sequence into the existing backbone. Gaps may be inserted where they help maximize good interactions. Good interactions are generally determined by pairwise potential energy functions.
The quality of the potential energy function or forcefield and the quality and thoroughness of the backbone database are both determining factors in how effective threading can be. This is an alternative to more physical methods of solving the protein folding problem. Rather than using laws of physical interactions, this method uses the extensive databases of sequence and structure to knowledge based structure prediction.