Structure not changing after pose.set_phi

Hi, I am trying to implement the Generalized Simulated Annealing routine (Tsalis et al, 1996) and apply it on folding processes.

To do so, I code the mover w.r.t. dihedral angles (each step phi and psi sufer a small, guided perturbation). I am using the pose.set_phi and .set_psi methods to modify pose coordinates. 

The problem:

When I do apply the Mover, it wont modify the pose at all. I alread tried several approaches to overload the Mover class, but none of them work. The last atempt will be down here. I also tried modifying the “Basic folding” example to move with all phi and psi at once, but it didn’t worked as well. It, strangely, only works if, at each step, I change only phi and psi of one residue (in the same fashion as the basic folding example).

The code:

class GSAMover(pyrosetta.rosetta.protocols.moves.Mover):

"""

GSA Mover:

"""

def __init__(self, **kwargs):

pyrosetta.rosetta.protocols.moves.Mover.__init__(self)



# . . . Just poping parameters from kwargs



def fetch_dihedrals(self, pose):

''' Get protein's dihedral angles '''

seqLen = len(pose.sequence())

dih = np.empty((seqLen,2))

for i in range(seqLen):

dih[i,] = np.asarray([pose.phi(i+1),pose.psi(i+1)])

return dih



def modify_type1(self, pose):

dihs = self.fetch_dihedrals(pose)

# x + g(r)

dihs = np.add(dihs, np.degrees(self.g(self.r_i)))

for i,dih in enumerate(dihs[1:]):

pose.set_phi(i+2,dih[0])

pose.set_psi(i+2,dih[1])

return pose



def apply(self, pose):

"""Applies move to pose."""



self.best_pose = Pose()

self.previous_pose = Pose()

#pose = pose.get()

for self.epoch in range(self.max_iter):

if self.epoch == 0:

self.best_pose.assign(pose)

self.previous_pose.assign(pose)

self.pose_zero = Pose()

self.pose_zero.assign(pose)



self.pose_modified = Pose()

self.pose_modified.assign(self.modifyPose(pose))



pose.assign(self.decisionCriteria(self.pose_modified,

self.pose_zero))



if self.verbose:

print('** Runing iteration: {}'.format(self.epoch))



if self.score(pose) < self.score(self.best_pose):

self.best_pose.assign(pose)



self.log.append(self.tqt)

self.log.append(self.g(self.r_i))

self.previous_pose.assign(pose)

Are there any other steps/methods required for me to move the pose?

Thank you for your attention.