Notebook

Rosetta Constraint API¶

@Author: 吴炜坤

@email: weikun.wu@xtalpi.com

在前面的章节中，我们介绍了在Rosetta中如何定义需要的约束文件。这些文件的定义和约束都是手动的。在本章节中，我们将会介绍:

如何从file读取constraint至pose中如何使用constraint生成器自动生成约束如何管理Pose中的约束

一、从几何约束文件读取约束¶

这里我们将使用ROSETTA中的constraint file（限制文件）。举例，我们当前对20号和6号Ca原子进行距离的约束，目标约束距离为9埃。

constraint file的内容如下:

AtomPair CA 20 CA 6 LINEAR_PENALTY 9.0 0 0 1.0

In [1]:

# 在1ubq_clean.pdb的例子上施加原子对的限制

from pyrosetta import init, create_score_function, pose_from_pdb
from pyrosetta.rosetta.core.scoring import ScoreType
init()

PyRosetta-4 2021 [Rosetta PyRosetta4.conda.mac.cxx11thread.serialization.python37.Release 2021.26+release.b308454c455dd04f6824cc8b23e54bbb9be2cdd7 2021-07-02T13:01:54] retrieved from: http://www.pyrosetta.org
(C) Copyright Rosetta Commons Member Institutions. Created in JHU by Sergey Lyskov and PyRosetta Team.
core.init: {0} Checking for fconfig files in pwd and ./rosetta/flags
core.init: {0} Rosetta version: PyRosetta4.conda.mac.cxx11thread.serialization.python37.Release r288 2021.26+release.b308454c455 b308454c455dd04f6824cc8b23e54bbb9be2cdd7 http://www.pyrosetta.org 2021-07-02T13:01:54
core.init: {0} command: PyRosetta -ex1 -ex2aro -database /opt/miniconda3/lib/python3.7/site-packages/pyrosetta/database
basic.random.init_random_generator: {0} 'RNG device' seed mode, using '/dev/urandom', seed=-1705239655 seed_offset=0 real_seed=-1705239655 thread_index=0
basic.random.init_random_generator: {0} RandomGenerator:init: Normal mode, seed=-1705239655 RG_type=mt19937

In [2]:

# 初始化REF2015的打分函数对象:
scorefxn = create_score_function('ref2015')

# 激活相关惩罚项的打分函数权重:
scorefxn.set_weight(ScoreType.atom_pair_constraint, 1.0) 

core.scoring.etable: {0} Starting energy table calculation
core.scoring.etable: {0} smooth_etable: changing atr/rep split to bottom of energy well
core.scoring.etable: {0} smooth_etable: spline smoothing lj etables (maxdis = 6)
core.scoring.etable: {0} smooth_etable: spline smoothing solvation etables (max_dis = 6)
core.scoring.etable: {0} Finished calculating energy tables.
basic.io.database: {0} Database file opened: scoring/score_functions/hbonds/ref2015_params/HBPoly1D.csv
basic.io.database: {0} Database file opened: scoring/score_functions/hbonds/ref2015_params/HBFadeIntervals.csv
basic.io.database: {0} Database file opened: scoring/score_functions/hbonds/ref2015_params/HBEval.csv
basic.io.database: {0} Database file opened: scoring/score_functions/hbonds/ref2015_params/DonStrength.csv
basic.io.database: {0} Database file opened: scoring/score_functions/hbonds/ref2015_params/AccStrength.csv
core.chemical.GlobalResidueTypeSet: {0} Finished initializing fa_standard residue type set.  Created 984 residue types
core.chemical.GlobalResidueTypeSet: {0} Total time to initialize 0.684347 seconds.
basic.io.database: {0} Database file opened: scoring/score_functions/rama/fd/all.ramaProb
basic.io.database: {0} Database file opened: scoring/score_functions/rama/fd/prepro.ramaProb
basic.io.database: {0} Database file opened: scoring/score_functions/omega/omega_ppdep.all.txt
basic.io.database: {0} Database file opened: scoring/score_functions/omega/omega_ppdep.gly.txt
basic.io.database: {0} Database file opened: scoring/score_functions/omega/omega_ppdep.pro.txt
basic.io.database: {0} Database file opened: scoring/score_functions/omega/omega_ppdep.valile.txt
basic.io.database: {0} Database file opened: scoring/score_functions/P_AA_pp/P_AA
basic.io.database: {0} Database file opened: scoring/score_functions/P_AA_pp/P_AA_n
core.scoring.P_AA: {0} shapovalov_lib::shap_p_aa_pp_smooth_level of 1( aka low_smooth ) got activated.
basic.io.database: {0} Database file opened: scoring/score_functions/P_AA_pp/shapovalov/10deg/kappa131/a20.prop

从文件读取约束信息主要依靠ConstraintSetMover的作用:

In [3]:

from pyrosetta.rosetta.protocols.constraint_movers import ConstraintSetMover
cst_set = ConstraintSetMover()
cst_set.add_constraints(False) # True=在原有限制基础上额外再添加限制，False= 从文件中读取并覆盖所有的限制。
cst_set.constraint_file('./data/constraint_file')

In [4]:

# load pose from 1ubq_clean.pdb
pose = pose_from_pdb("./data/1ubq_clean.pdb")
cst_set.apply(pose)

core.import_pose.import_pose: {0} File './data/1ubq_clean.pdb' automatically determined to be of type PDB
core.scoring.constraints.ConstraintsIO: {0} read constraints from ./data/constraint_file
core.scoring.constraints.ConstraintsIO: {0} Read in 1 constraints

In [5]:

# 打印查看当前pose所有设定的constraint
print(pose.constraint_set())

ResiduePairConstraints: total: 20   plotting active...
ResiduePairConstraints (6,20)
AtomPairConstraint (2,20-2,6)
         r      func     dfunc dfunc_est
         2     7.000    -1.000    -1.000
       2.5     6.500    -1.000    -1.000
         3     6.000    -1.000    -1.000
       3.5     5.500    -1.000    -1.000
         4     5.000    -1.000    -1.000
       4.5     4.500    -1.000    -1.000
         5     4.000    -1.000    -1.000
       5.5     3.500    -1.000    -1.000
         6     3.000    -1.000    -1.000
       6.5     2.500    -1.000    -1.000
         7     2.000    -1.000    -1.000
       7.5     1.500    -1.000    -1.000
         8     1.000    -1.000    -1.000
       8.5     0.500    -1.000    -1.000
         9     0.000     0.000     0.000
       9.5     0.500     1.000     1.000
        10     1.000     1.000     1.000
      10.5     1.500     1.000     1.000
        11     2.000     1.000     1.000
      11.5     2.500     1.000     1.000
        12     3.000     1.000     1.000
      12.5     3.500     1.000     1.000
        13     4.000     1.000     1.000
      13.5     4.500     1.000     1.000
        14     5.000     1.000     1.000
      14.5     5.500     1.000     1.000
        15     6.000     1.000     1.000
      15.5     6.500     1.000     1.000
        16     7.000     1.000     1.000
      16.5     7.500     1.000     1.000
        17     8.000     1.000     1.000
      17.5     8.500     1.000     1.000
        18     9.000     1.000     1.000
      18.5     9.500     1.000     1.000
        19    10.000     1.000     1.000
      19.5    10.500     1.000     1.000
        20    11.000     1.000     1.000

ResiduePairConstraints (20,6)
AtomPairConstraint (2,20-2,6)
         r      func     dfunc dfunc_est
         2     7.000    -1.000    -1.000
       2.5     6.500    -1.000    -1.000
         3     6.000    -1.000    -1.000
       3.5     5.500    -1.000    -1.000
         4     5.000    -1.000    -1.000
       4.5     4.500    -1.000    -1.000
         5     4.000    -1.000    -1.000
       5.5     3.500    -1.000    -1.000
         6     3.000    -1.000    -1.000
       6.5     2.500    -1.000    -1.000
         7     2.000    -1.000    -1.000
       7.5     1.500    -1.000    -1.000
         8     1.000    -1.000    -1.000
       8.5     0.500    -1.000    -1.000
         9     0.000     0.000     0.000
       9.5     0.500     1.000     1.000
        10     1.000     1.000     1.000
      10.5     1.500     1.000     1.000
        11     2.000     1.000     1.000
      11.5     2.500     1.000     1.000
        12     3.000     1.000     1.000
      12.5     3.500     1.000     1.000
        13     4.000     1.000     1.000
      13.5     4.500     1.000     1.000
        14     5.000     1.000     1.000
      14.5     5.500     1.000     1.000
        15     6.000     1.000     1.000
      15.5     6.500     1.000     1.000
        16     7.000     1.000     1.000
      16.5     7.500     1.000     1.000
        17     8.000     1.000     1.000
      17.5     8.500     1.000     1.000
        18     9.000     1.000     1.000
      18.5     9.500     1.000     1.000
        19    10.000     1.000     1.000
      19.5    10.500     1.000     1.000
        20    11.000     1.000     1.000

IntraResidueConstraints: total: 0 showing active...
NonResiduePairConstraints: total: 0 showing active...

从这个简单的例子中可以看到，使用“限制文件”对pose进行限制的方法理论上适用于所有的场景。因为只要我们按照Rosetta的要求编写所有的限制规则于限制文件中，这种方法就可以最大化用户自由度，但是缺点是使用“限制文件”事先需要对限制文件的编写规则有一定的了解和把握。

二、从序列约束文件读取约束¶

主要通过AddCompositionConstraintMover，AddNetChargeConstraintMover两个函数来进行约束的添加。

2.1 AddCompositionConstraintMover¶

给氨基酸的组成比例添加限制。开始使用前，需要准备*.comp限制文件，控制如何限制氨基酸的比例。

In [6]:

from pyrosetta.rosetta.core.scoring import ScoreType
score = create_score_function('ref2015')
score.set_weight(ScoreType.aa_composition, 1.0) # reweight score

from pyrosetta.rosetta.protocols.aa_composition import AddCompositionConstraintMover
add_comp_cst = AddCompositionConstraintMover()
add_comp_cst.create_constraint_from_file('./data/aa_comp_constraint_file.comp')
# add_comp_cst.add_residue_selector(selector) # 输入pose限制的范围, 如果缺省，那么将全pose限制氨基酸组成。
add_comp_cst.apply(pose)

protocols.aa_composition.AddCompositionConstraintMover: {0} Initialized AACompositionConstraint object from file ./data/aa_comp_constraint_file.comp.

2.2 AddNetChargeConstraintMover¶

添加净电荷限制Mover,开始使用前，需要准备.charge限制文件。

In [7]:

from pyrosetta.rosetta.core.scoring import ScoreType
score = create_score_function('ref2015')
score.set_weight(ScoreType.netcharge, 1.0) # reweight score

from pyrosetta.rosetta.protocols.aa_composition import AddNetChargeConstraintMover
add_net_charge_cst = AddNetChargeConstraintMover()
add_net_charge_cst.create_constraint_from_file('./data/charge_comp_constraint_file.charge')
# add_net_charge_cst.add_residue_selector(selector) # 输入pose限制的范围, 如果缺省，那么将全pose限制氨基酸组成。
add_net_charge_cst.apply(pose)

core.scoring.netcharge_energy.NetChargeEnergySetup: {0} Reading netcharge scoring term setup data from ./data/charge_comp_constraint_file.charge.
protocols.aa_composition.AddNetChargeConstraintMover: {0} Initialized NetChargeConstraint object from file ./data/charge_comp_constraint_file.charge.

2.3 ResidueTypeConstraintMover¶

在Design时倾向于奖励某类氨基酸的能量得分。如当突变成Gly时 ddg=-1，突变成A的ddg=-1，那么经过ALA bonus约束(weight=2.0). 最后氨基酸序列输出时，选择了突变成A氨基酸的ddg为-2。这是一种直接在packer中奖励residue序列得分全局的方法。

In [8]:

from pyrosetta.rosetta.protocols.constraint_movers import ResidueTypeConstraintMover
aa_type_cst = ResidueTypeConstraintMover()
aa_type_cst.set_AA_name3('ALA')
aa_type_cst.set_favor_bonus(2.0) #权重
aa_type_cst.apply(pose)

三、使用Constraint Generators生成约束¶

Constraint Generators不同于file文件系统，生成器可以自动地根据用户输入的参数进行约束参数生成，十分方便。

备注: 以下内容可能涉及ResidueSelector，介绍详见第五章节。

以下我们将介绍几种常用的自动生成器的使用方法:

3.1 AddConstraints¶

AddConstraintsMover可以使用多个constraint generator来生成一些列的限制，并且将他们添加到Pose中。

后面将在例子中进行使用说明。

3.2 AtomPairConstraintGenerator¶

在一个selector子集内部生成遍历的Pair-wise的原子距离约束。

有两种使用模式:

模式1: 在一个子集中生成距离限制

In [9]:

from pyrosetta.rosetta.protocols.simple_moves import VirtualRootMover

# load pose from 1ubq_clean.pdb
pose = pose_from_pdb("./data/1ubq_clean.pdb")

# 添加虚拟的root
vr_mover = VirtualRootMover()
vr_mover.apply(pose)

core.import_pose.import_pose: {0} File './data/1ubq_clean.pdb' automatically determined to be of type PDB

In [10]:

# Score reweight
score = create_score_function('ref2015')
score.set_weight(ScoreType.atom_pair_constraint, 1.0) # reweight constraint score

# 定义生成器
from pyrosetta.rosetta.protocols.constraint_generator import AtomPairConstraintGenerator
atompair_cst = AtomPairConstraintGenerator()

# 定义残基选择器
from pyrosetta.rosetta.core.select.residue_selector import ChainSelector
chain_selector = ChainSelector(1)

# 结构设置
atompair_cst.set_residue_selector(chain_selector)

# 参数设置
atompair_cst.set_ca_only(True)
atompair_cst.set_max_distance(8) # 超出这个距离不添加限制；
atompair_cst.set_min_seq_sep(6)  # 如果序列氨基酸号大于该值才添加限制。
atompair_cst.set_sd(1.0) # distance constraint的标准误;
atompair_cst.set_use_harmonic_function(True) # use harmonic function instead of SOG function

# add AtomPairConstraintGenerator to pose;
from pyrosetta.rosetta.protocols.constraint_generator import AddConstraints
add_cst = AddConstraints()
add_cst.add_generator(atompair_cst)
add_cst.apply(pose)

protocols.constraint_generator.AddConstraints: {0} Adding 116 constraints generated by ConstraintGenerator named unnamed_constraint_generator

模式2: 生成两个子集之间的距离限制

In [11]:

from pyrosetta.rosetta.protocols.simple_moves import VirtualRootMover

# load pose from 1ubq_clean.pdb
pose = pose_from_pdb("./data/1ubq_clean.pdb")

# 添加虚拟的root
vr_mover = VirtualRootMover()
vr_mover.apply(pose)

core.import_pose.import_pose: {0} File './data/1ubq_clean.pdb' automatically determined to be of type PDB

In [12]:

# Score reweight
score = create_score_function('ref2015')
score.set_weight(ScoreType.atom_pair_constraint, 1.0) # reweight constraint score

# 定义生成器
from pyrosetta.rosetta.protocols.constraint_generator import AtomPairConstraintGenerator
atompair_cst = AtomPairConstraintGenerator()

# 定义残基选择器
from pyrosetta.rosetta.core.select.residue_selector import ResidueIndexSelector
res1_selector = ResidueIndexSelector('1-10')
res2_selector = ResidueIndexSelector('20-30')

# 结构设置
atompair_cst.set_residue_selector(res1_selector)
atompair_cst.set_secondary_residue_selector(res2_selector) #生成两个selector之间的距离限制时需要，使用后，min_seq_sep将不起效。

# 参数设置
atompair_cst.set_ca_only(True)
atompair_cst.set_max_distance(12) # 超出这个距离不添加限制；
atompair_cst.set_sd(1.0) # distance constraint的标准误;
atompair_cst.set_use_harmonic_function(True) # use harmonic function instead of SOG function

# add AtomPairConstraintGenerator to pose;
from pyrosetta.rosetta.protocols.constraint_generator import AddConstraints
add_cst = AddConstraints()
add_cst.add_generator(atompair_cst)
add_cst.apply(pose)

protocols.constraint_generator.AddConstraints: {0} Adding 14 constraints generated by ConstraintGenerator named unnamed_constraint_generator

3.3 CoordinateConstraintGenerator¶

在一个Selector子集内部生成坐标限制;

注意: 需要使用pyrosetta.rosetta.protocols.simple_moves.VirtualRootMover加一个虚拟的root.

In [13]:

from pyrosetta.rosetta.protocols.simple_moves import VirtualRootMover

# load pose from 1ubq_clean.pdb
pose = pose_from_pdb("./data/1ubq_clean.pdb")

# 添加虚拟的root
vr_mover = VirtualRootMover()
vr_mover.apply(pose)

core.import_pose.import_pose: {0} File './data/1ubq_clean.pdb' automatically determined to be of type PDB

In [14]:

# Score reweight
score = create_score_function('ref2015')
score.set_weight(ScoreType.coordinate_constraint, 1.0) # reweight constraint score

# 定义残基选择器
from pyrosetta.rosetta.core.select.residue_selector import ResidueIndexSelector
res1_selector = ResidueIndexSelector('1-10')


# 定义坐标约束生成器
from pyrosetta.rosetta.protocols.constraint_generator import CoordinateConstraintGenerator
coordinate_cst = CoordinateConstraintGenerator()
coordinate_cst.set_ca_only(False)
coordinate_cst.set_reference_pose(pose)
coordinate_cst.set_residue_selector(res1_selector)
coordinate_cst.set_sidechain(False) # True = all重原子，False = 骨架重原子

# 约束函数设定
coordinate_cst.set_sd(1.0) # strength/deviation e.g. -relax:coord_cst_stdev

# add AtomPairConstraintGenerator to pose;
from pyrosetta.rosetta.protocols.constraint_generator import AddConstraints
add_cst = AddConstraints()
add_cst.add_generator(coordinate_cst)
add_cst.apply(pose)

protocols.constraint_generator.AddConstraints: {0} Adding 40 constraints generated by ConstraintGenerator named unnamed_constraint_generator

3.4 DihedralConstraintGenerator¶

基于骨架二面角(phi or psi or omega)自动生成约束的生成器, 默认使用CircularHarmonic函数。

In [22]:

from pyrosetta.rosetta.protocols.simple_moves import VirtualRootMover

# load pose from 1ubq_clean.pdb
pose = pose_from_pdb("./data/1ubq_clean.pdb")

# 添加虚拟的root
vr_mover = VirtualRootMover()
vr_mover.apply(pose)

core.import_pose.import_pose: {0} File './data/1ubq_clean.pdb' automatically determined to be of type PDB

In [24]:

from pyrosetta.rosetta.protocols.constraint_generator import DihedralConstraintGenerator
from pyrosetta.rosetta.core.id import MainchainTorsionType

# Score reweight
score = create_score_function('ref2015')
score.set_weight(ScoreType.dihedral_constraint, 1.0) # reweight score

# 
dh_cst = DihedralConstraintGenerator()
# dh_cst.set_residue_selector()  # 设置选择器
dh_cst.set_sd_degree(7.0)  # 默认允许的最大变化角度
dh_cst.set_torsion_type(MainchainTorsionType.phi_dihedral)  # phi_dihedral,psi_dihedral,omega_dihedral

# add DihedralConstraintGenerator to pose;
from pyrosetta.rosetta.protocols.constraint_generator import AddConstraints
add_cst = AddConstraints()
add_cst.add_generator(termin_cst)
add_cst.apply(pose)

protocols.constraint_generator.TerminiConstraintGenerator: {0} Constraining atoms  atomno= 2 rsd= 1  and  atomno= 2 rsd= 76 , min_distance=8 max_distance=20
protocols.constraint_generator.AddConstraints: {0} Adding 1 constraints generated by ConstraintGenerator named unnamed_constraint_generator

3.5 TerminiConstraintGenerator¶

生成N端和C端之间原子的限制参数。从PyRosetta API中看，仅仅是可以设置简单的距离限制。

In [15]:

from pyrosetta.rosetta.protocols.simple_moves import VirtualRootMover

# load pose from 1ubq_clean.pdb
pose = pose_from_pdb("./data/1ubq_clean.pdb")

# 添加虚拟的root
vr_mover = VirtualRootMover()
vr_mover.apply(pose)

core.import_pose.import_pose: {0} File './data/1ubq_clean.pdb' automatically determined to be of type PDB

In [16]:

# Score reweight
score = create_score_function('ref2015')
score.set_weight(ScoreType.atom_pair_constraint, 1.0) # reweight score

# 
from pyrosetta.rosetta.protocols.constraint_generator import TerminiConstraintGenerator
termin_cst = TerminiConstraintGenerator()
termin_cst.set_min_distance(8)
termin_cst.set_max_distance(20)
termin_cst.set_sd(1.0)

# add TerminiConstraintGenerator to pose;
from pyrosetta.rosetta.protocols.constraint_generator import AddConstraints
add_cst = AddConstraints()
add_cst.add_generator(termin_cst)
add_cst.apply(pose)

protocols.constraint_generator.TerminiConstraintGenerator: {0} Constraining atoms  atomno= 2 rsd= 1  and  atomno= 2 rsd= 76 , min_distance=8 max_distance=20
protocols.constraint_generator.AddConstraints: {0} Adding 1 constraints generated by ConstraintGenerator named unnamed_constraint_generator

3.6 AutomaticSheetConstraintGenerator¶

自动根据Pose中的beta sheet结构匹配，生成sheet间氢键的配对约束。

In [17]:

from pyrosetta.rosetta.protocols.fold_from_loops.constraint_generator import AutomaticSheetConstraintGenerator
# Score reweight
score = create_score_function('ref2015')
score.set_weight(ScoreType.atom_pair_constraint, 1.0) # reweight score

sheet_g = AutomaticSheetConstraintGenerator()
sheet_g.distance(12)
sheet_g.sd(1.0)

# add AutomaticSheetConstraintGenerator to pose;
from pyrosetta.rosetta.protocols.constraint_generator import AddConstraints
add_cst = AddConstraints()
add_cst.add_generator(termin_cst)
add_cst.apply(pose)

protocols.constraint_generator.TerminiConstraintGenerator: {0} Constraining atoms  atomno= 2 rsd= 1  and  atomno= 2 rsd= 76 , min_distance=8 max_distance=20
protocols.constraint_generator.AddConstraints: {0} Adding 1 constraints generated by ConstraintGenerator named unnamed_constraint_generator

四、清除Pose中的约束限制的方法¶

4.1 ReleaseConstraintFromResidueMover¶

从Pose中清除selector中的氨基酸限制。

In [18]:

from pyrosetta.rosetta.protocols.fold_from_loops.movers import ReleaseConstraintFromResidueMover
from pyrosetta.rosetta.core.select.residue_selector import ChainSelector

# 定义残基选择器
chain_selector = ChainSelector(1)

# 清除约束
clean_selector_cst = ReleaseConstraintFromResidueMover(chain_selector)
clean_selector_cst.apply(pose)

4.2 RemoveConstraints¶

从Pose中清除由constraint generator生成的限制。

In [19]:

# add TerminiConstraintGenerator to pose again;
from pyrosetta.rosetta.protocols.constraint_generator import AddConstraints
add_cst = AddConstraints()
add_cst.add_generator(termin_cst)
add_cst.apply(pose)


from pyrosetta.rosetta.protocols.constraint_generator import RemoveConstraints
rm_cst = RemoveConstraints()
rm_cst.add_generator(termin_cst)  # termin_cst为之前定义的生成器;
rm_cst.apply(pose)

protocols.constraint_generator.TerminiConstraintGenerator: {0} Constraining atoms  atomno= 2 rsd= 1  and  atomno= 2 rsd= 76 , min_distance=8 max_distance=20
protocols.constraint_generator.AddConstraints: {0} Adding 1 constraints generated by ConstraintGenerator named unnamed_constraint_generator
protocols.constraint_generator.RemoveConstraints: {0} Removing 1 constraints from pose generated by unnamed_constraint_generator

4.3 ClearConstraintsMover¶

从Pose中清除所有的限制。

In [20]:

from pyrosetta.rosetta.protocols.constraint_movers import ClearConstraintsMover
clear_cst = ClearConstraintsMover()
clear_cst.apply(pose)

4.4 ClearCompositionConstraintsMover¶

从Pose中清除sequence constraint/net charge constraint.

In [21]:

from pyrosetta.rosetta.protocols.aa_composition import ClearCompositionConstraintsMover
clear_seq_cst = ClearCompositionConstraintsMover()
clear_seq_cst.apply(pose)

protocols.aa_composition.ClearCompositionConstraintsMover: {0} Removed all sequence constraints from the pose.

In [ ]: