Laboratory for Structural Bioinformatics

Although both Rosetta and RosettaX can generate models that can be used successfully for molecular replacement, there are cases when both methods fail. After examining those failed cases, we found that one of the primary reason is the models generated are not sufficiently accurate to enable molecular replacement success. To address this problem, we have developed a model rebuilding method that focuses the conformational search on residues with large errors in order improve the overall model quality to a level that is sufficient for molecular replacement. The residue coordinates error is predicted by an average pairwise residue distance score (AP-RDS). Our method is termed Morpheus (publication). On a test dataset of 10 difficult structures, Morpheus has succeeded in 7 while the latest version of Rosetta3.2 has succeeded in 2 cases.