< Back to Article

Heuristic energy-based cyclic peptide design

Fig 3

Stability analysis of 7-residue designs.

(a) Correlation plot between values calculated by Rosetta simple_cycpep_predict and by our filtering method. (b) value distributions within different energy bins (kcal/mol). Design counts are labeled on top of the bars. (c) Energy landscape comparisons for three cases that have noticeable differences in values calculated by the two methods. Left, more extensive sampling of wells further from the designed state by the filtering method results in a lower computed value. Middle: more extensive sampling close to the designed state by the filtering method identifies a deeper minimum, raising the value. Right: more extensive sampling by the filtering method allows exploration in a low-RMSD region missed entirely by Rosetta, identifying energy wells and raising the value. (d) Top 7-residue designs () demonstrating smaller backbone root-mean-square radii with H-bond intersections. Representative designs having 0, 1, 2, 3, 5, and 6 H-bond intersections are drawn, along with their H-bond networks where L- and D-amino acids are specified and arrows point from amide proton to carbonyl oxygen. The designed sequences are written in one-letter codes, with uppercase for L-amino acids, and lowercase for D-amino acids.

doi: https://doi.org/10.1371/journal.pcbi.1012290.g003