PyRosetta (8): Relax

Overview

The word “relax” is borrowed from thermodynamics; it refers to the process in which a system gradually recovers to a balanced or equilibrium state from a unbalanced state.

The input pose is considered as a “unbalanced” system, and through the “relax” process, PyRosetta aims to find a “balanced” (globally minimum) system.

Relax is the main protocol for simple all-atom refinement of structures in the Rosetta force-field. Relax does not do extensive refinement and only searches the local conformational space around the starting structure. Relax is thus often used in conjunction with more aggressive sampling protocols like fragment assembly (abinitio) and loop modelling. To evaluate different conformations based on their Rosetta all-atom score one usually has to apply relax.

It can also read centroid models, in which case it will convert the model into a full-atom model and pack the sidechains. Relax does not carry out any extensive refinement and only searches the local conformational space neighbourhood.

It is further advisable to apply relax only to previously idealized structures. Idealization avoids that score differences arise due to non-ideal geometry (e.g., at the position of former chain-breaks introduced during an aggressive sampling stage and removed by loop closing).

https://docs.rosettacommons.org/docs/latest/application_documentation/structure_prediction/relax

ClassicRelax

Depreciated. Small move + shear move.

FastRelax(default)

5 cycles of {[packing + minimization] * n_ramping}.

Packing: optimizing side chain conformations without editing side chains.

Minimization: find local energy minima.

n_ramping: steps of ramping up fa_rep (in each single cycle). In each of the 5 cycles, the repulsion energy (fa_rep) are firstly set to a small number to allow exploration, and then it gradually increases to a realistic value in n-step ramping. The final stages of the ramp enforce the full, realistic energy function, ensuring that the resulting structure has physically correct packing and geometry.

Pseudo code:

# Assume that we have a `Pose` object called `my_pose`
for i in range(0,5): # "5" by default, can be modified
    fa_rep = 0.1 * fa_rep
    for j in range(1, n_ramp):
        fa_rep = Increase(fa_rep)
        Packing(my_pose)
        Minimizing(my_pose)

CentroidRelax(for massive screening)

1. It uses centroid score functions
2. It ramps up various energy terms while minimizing the pose
3. It is rough but fast, thus is suitable for high-throughput structure design
   (as a pre-processing & evaluation for further FastRelax).