All-atom energy vs. RMSD plots for <i>de novo</i> modeling of the four puzzles and for optimizing experimental (“native”) conformations.

<p>Panels correspond exactly to panels in <a href="http://www.plosone.org/article/info:doi/10.1371/journal.pone.0020044#pone-0020044-g001" target="_blank">Fig. 1</a>. In protein cases (A)-(C), the default Rosetta all-atom energy function for <i>de novo</i> protein modeling (score12) is plotted against Cα RMSD. In the RNA case (D), the FARFAR energy function (which contains torsional terms for RNA, an orientation-dependent solvation function, and a carbon-hydrogen-bond model <a href="http://www.plosone.org/article/info:doi/10.1371/journal.pone.0020044#pone.0020044-Das4" target="_blank">[19]</a>) is plotted against all-heavy-atom RMSD. The conformational sampling algorithms (ABRELAX, SWA, etc.) used in the runs are denoted in the figure and described in detail in Methods.