Questions of Cartesian_ddG calculation in PyRosetta

[Anonymous]

Dear All,

I am new to PyRosetta and I am trying to calculate the ddG after point-mutation. The calculation logic is like the following:

1) clean the pdb file and use cartesian relax to minimize its energy

2) perform the point mutation with mutate_residue() function

3) calculate the score before and after the mutation to gain the ddG

However, the ddG data I got was always positive values and have large differences from the benchmark.

Can someone give me some help? Which part of my code is wrong? Thank you guys so much for your input and suggestions!!

from pyrosetta import *

from pyrosetta.rosetta import *

from pyrosetta.toolbox import *

init("-ignore_unrecognized_res 1 -ex1 -ex2 -flip_HNQ -relax:cartesian -nstruct 200 -crystal_refine -optimization:default_max_cycles 200")





testPose= Pose()

testPose = pose_from_pdb("1BNI.clean.pdb")



from pyrosetta.rosetta.protocols.relax import FastRelax

scorefxn = pyrosetta.create_score_function("ref2015_cart")

relax = pyrosetta.rosetta.protocols.relax.FastRelax()

relax.constrain_relax_to_start_coords(True)

relax.coord_constrain_sidechains(True)

relax.ramp_down_constraints(False)

relax.cartesian(True)

#relax.min_type("dfpmin_armijo_nonmonotone")

relax.min_type("lbfgs_armijo_nonmonotone")#for non-Cartesian scorefunctions use'dfpmin_armijo_nonmonotone'

relax.set_scorefxn(scorefxn)

s0=scorefxn(testPose)

relax.apply(testPose)

s1=scorefxn(testPose)

testPose.dump_pdb('1BNI.relax.pdb')

AA=

ddG=[]

mp=Pose()

for i in AA:

mp.assign(testPose)

mutate_residue(mp,52,i)

s2=scorefxn(mp)

dg=s2-s1

ddG.append(dg)



print(ddG)

 

 Best regards

[Anonymous]

Hi,

I would relax or minimise after the mutation. Not sure `mutate_residue` does this allready. I would have a flexible backbone around the mutation (i.e. 8 Angs), but keep everything else fixed. 

Here is one implementation of ddg https://github.com/Kortemme-Lab/ddg

And another: https://www.rosettacommons.org/docs/latest/cartesian-ddG

And the newest one in the rosetta_scripts directory (not sure it’s been released yet):

rosetta_scripts_scripts/scripts/public/stabilize_proteins_pm_mc/

Best,

Ajasja

PS: also what’s the diffrence between s0 and s1? Does relax take into account the coord constraints? I would use two diffrent scoring functions one for relax and one for scoring, so there would be no doubt if constraints are on.

[Anonymous]

Hi Ajasja,

Thank you so so much for your kindly reply. First of all, I would like to say sorry that I am new to both Rosetta and PyRosetta so you may find my questions are a bit silly. I checked the reference you sent to me but seems that both of them are only suitable for Rosetta, while I want to do the job in PyRosetta.

Regarding the questions in your PS section. I found that the energy before and after the relax was different. s0 is a positive number while s1 is a negative number. I don’t know whether in my python code the relax takes into account the coord constraints, but according to the official tutorial the command “relax.constrain_relax_to_start_coords(True)” helps me do it? Also, may I ask you what do you mean by “using two different scoring functions?”, do you mean that I can set up a scorefxn1=”ref2015_cart” for cartesian_relax, while setting up another scorefxn2=’ref2015′ for the scoring calculation?

For your suggestion, May I ask how you keep everything else fixed while having a flexible backbone around the mutation? Do I have to do the minimization for the mutated pose? The “mutate_residue()” function is attached below, does the “Packer“ help me do the job already?

Really sorry to bother you again and again, looking forward to hearing from you!

Best,

Kai San

def mutate_residue(

pack_or_pose, mutant_position, mutant_aa, pack_radius=0., pack_scorefxn=None

):

"""Replace the residue at a single position in a Pose with a new amino acid

and repack any residues within user-defined radius of selected residue's

center using.

Args:

pack_or_pose (pyrosetta.rosetta.core.pose.Pose OR pyrosetta.distributed.packed_pose.PackedPose):

the `Pose` instance to use.

mutant_position (int): Pose-numbered position of the residue to mutate.

mutant_aa (str): The single letter name for the desired amino acid.

pack_radius (float): Radius used to define neighboring residues.

pack_scorefxn (pyrosetta.ScoreFunction): `ScoreFunction` to use when repacking the `Pose`.

Defaults to the standard `ScoreFunction`.

"""

import pyrosetta

import pyrosetta.distributed.packed_pose as packed_pose



wpose = packed_pose.to_pose(pack_or_pose)



if not wpose.is_fullatom():

raise IOError("mutate_residue only works with fullatom poses")



# create a standard scorefxn by default

if not pack_scorefxn:

pack_scorefxn = pyrosetta.get_score_function()



# the numbers 1-20 correspond individually to the 20 proteogenic amino acids

from pyrosetta.rosetta.core.chemical import aa_from_oneletter_code



mutant_aa = int(aa_from_oneletter_code(mutant_aa))

aa_bool = pyrosetta.Vector1([aa == mutant_aa for aa in range(1, 21)])



# mutation is performed by using a PackerTask with only the mutant

# amino acid available during design

task = pyrosetta.standard_packer_task(wpose)

task.nonconst_residue_task(mutant_position).restrict_absent_canonical_aas(aa_bool)



# prevent residues from packing by setting the per-residue "options" of the PackerTask

task = restrict_non_nbrs_from_repacking(wpose, mutant_position, task, pack_radius)



# apply the mutation and pack nearby residues

from pyrosetta.rosetta.protocols.minimization_packing import PackRotamersMover



packer = PackRotamersMover(pack_scorefxn, task)

packer.apply(wpose)

 

[Author]

Posts