Successful ab initio prediction of the tertiary structure of NK-lysin using multiple sequences and recognized supersecondary structural motifs.

A simple approach to protein tertiary structure prediction is described, based on the assembly of recognized supersecondary structural fragments taken from highly resolved protein structures by using a simulated annealing algorithm. The results of blind-testing this method on CASP2 target T0042 (pig NK-lysin) are presented. The predicted structure had a C alpha root-mean-square deviation of only 6.2 A from the experimental structure (and less than 5.0 A over the first 66 residues), and clearly had the correct fold when judged by using a number of objective measures. Despite the significant degree of success in this case, there is clearly much more development required before predictions with the accuracy of a good homology model can be made with this kind of approach.