In an attempt to simplify the general problem of protein folding, we have adopted a hierarchical approach. First, identify the location of secondary structure elements. Then, pack these pre-formed secondary structures to form an approximate tertiary structure. Finally, regularize and optimize the approximate tertiary structure to yield a detailed view of the folded protein. Interleukin-4 (IL-4) provides a recent example of our efforts at structure prediction. The structure was predicted in advance of the determination of the three-dimensional structure by Nuclear Magnetic Resonance (NMR) spectroscopy by Chris Dobson and colleagues at Oxford. This allows an unbiased comparison of prediction and experiment. In general, it is difficult to predict accurately the secondary structure of an arbitrary protein. When the folding class of a protein is known (a/a, a/b, b/b), the accuracy of secondary structure prediction algorithms improves. This is most obviously true for all helical (a/a) proteins. For interleukin-4, circular dichroism spectroscopy was used to establish the allhelical character of the molecule. Loop regions were located using a pattern-matching algorithm. Helical regions were identified using a combination of methods. The prediction was 87% correct and 93% correct, if a variation of one residue in the helix boundary is tolerated.
Showing the most recent 10 out of 508 publications
