Analysis of the predicted structures of domain 1 of protein g3 (T0030) and NK-lysin (T0042)

Structure predictions for two targets from the CASP2 meeting are presented and compared with the experimental structure. These predictions were made using a novel simplified flexible geometry representation of protein structure. The method uses potentials which mimic the physical forces involved in protein folding in a simplified representation of protein structure, and not by directly using data derived from statistical analyses of known protein structures. Additionally, the method is designed to work with a single protein sequence. The method was successful in generating reasonable protein-like structures, with mainly buried hydrophobic residues, exposed charges, and a good fraction of secondary structure. Specific details of the structure were remarkably close to the experimental structure. However, the overall fold in both cases was totally wrong. Some specific causes of this incorrect folding are suggested and a major improvement to the algorithm proposed.