Improved Rosetta protein structure prediction with customised fragments libraries based on structural class annotations

Taking advantage of the reliable performance of structural class prediction software, limitations of fragment-based methods are addressed by integrating structural constraints in their fragment selection process. Using Rosetta, the proposed pipeline is evaluated on 70 former CASP targets containing up to 150 amino acids. Using CATH-based structural class annotations, enhancement of structure prediction performance is highly significant in terms of both GDT_TS (at least +2.6, p-values < 0.0005) and RMSD (−0.4, p-values < 0.005).Further analysis also shows that methods relying on class-based fragments produce higher accuracy conformations: GDT_TS (up to 10% in average).