Figure 1.
Histograms of decoys’ energy distribution: comparison between iterations ,
of EdaFold and Rosetta for
(left) and
(right).
The minimization process of EdaFold is more efficient than Rosetta at iteration . The Estimation of Distribution Algorithm allows EdaFold to increase the performances from iteration
to iteration
.
Table 1.
Performance of EdaFold and Rosetta.
Figure 2.
Histogram of CARMSD to native decoy distribution: comparison between iterations ,
of EdaFold and Rosetta for
(left) and
(right).
EdaFold is able to guide the sampling towards native structure: the percentage of near-native models is higher at iteration .
Figure 3.
Decoys distribution as a function of CARMSD to native structure.
EdaFold (in blue) is able to generate a higher percentage of decoys at less than Å from native on
targets out of
.
Figure 4.
Histograms for performance criteria: best
and
average CARMSD to native and best percentage of models generated at less than
Å away from the native structure.
Figure 5.
1scj models: EdaFold’s best model (in green), Å away from native (in yellow), Rosetta’s best model (in blue),
Å away from native.
Table 2.
Molecular replacement with Phaser.
Figure 6.
Deceptive landscape: energy as a function of CARMSD to native structure on target for EdaFold and Rosetta.
Figure 7.
Suitable landscape: energy as a function of CARMSD to native structure on target for EdaFold and Rosetta.
Table 3.
EdaFold parameter settings during experiments.