Fig. 3 | Nature Communications

Fig. 3

From: De novo main-chain modeling for EM maps using MAINMAST

Fig. 3

Modeling results of the 30 experimental EM maps by MAINMAST. a Cα RMSD of the top scoring (solid circles) and the best RMSD model among the top 10 scoring models (empty circles) by MAINMAST in comparison with the Rosetta top scoring models. The refinement by Pulchra and MDFF were applied to the models. For Rosetta, results using 0.8 for the consensus fraction was used, because it showed better results than the default setting (Supplementary Fig. 3). The points above the frame indicate that Rosetta could not model these proteins while MAINMAST made full models at the RMSD values. b Comparison between MAINMAST and Rosetta (with a 0.8 consensus setting) in terms of coverage and precision of models. Coverage (precision) is defined as the fraction of Cα atoms in the native structure (the model) which are closer than 3.0 Å to any Cα atoms in the model (the native structure). c Comparison between the top scoring (Top 1) model and the best RMSD model among the top 10 scoring model for each of the 30 EM maps. d Comparison of MAINMAST models before and after the refinement by Pulchra and MDFF. Models before the refinement were selected by the threading score while the scoring function of MDFF was used after the refinement. e RMSD of the models (the best among the top 10 socring models) by MAINMAST after refinement relative to the map resolution

Back to article page