AbstractBackground. Large-scale computational prediction of protein structures represents a cost-effective alternative to empirical structure determination