Notebook

Pose Input/Output¶

@Author: 吴炜坤

@email：weikun.wu@xtalpi.com/weikunwu@163.com

一、蛋白质的结构¶

蛋白质是大型生物分子，它由一个或多个由α-氨基酸残基组成的长链条组成。α-氨基酸分子呈线性排列，相邻α-氨基酸残基的羧基和氨基通过肽键连接在一起。蛋白质的分子结构可划分为四级，以描述其不同的方面：

蛋白质一级结构：组成蛋白质多肽链的线性氨基酸序列。
蛋白质二级结构：依靠不同氨基酸之间的C=O和N-H基团间的氢键形成的稳定结构，主要为α螺旋和β折叠。
蛋白质三级结构：通过多个二级结构元素在三维空间的排列所形成的一个蛋白质分子的三维结构。
蛋白质四级结构：用于描述由不同多肽链（亚基）间相互作用形成具有功能的蛋白质复合物分子。

(图片来源: google图片搜索)

在整个第一章中，你即将会学习到以下的一些信息：

Pose & Structure IO： Pose的生成与输出
PDBinfo & Pose：处理PDB和Pose编号的方式
Atom、Residue & ResidueType：理解原子和残基的定义方式
Conformation & Protein Geometry：理解构象与自由度
Pose operating：基本的一些Pose相关的实践操作

二、Pose的构架¶

因此如果要在计算机中建立一个蛋白质的结构模型，就清楚地描述每一个原子的信息。在Rosetta中，Pose是管理蛋白质信息的中心，可以描述蛋白质一到四级结构所有的信息。而且这些信息是分层管理的比如:

Conformation: 负责管理原子类型(AtomType)、氨基酸类型(ResidueType)、氨基酸的原子坐标(xyz)、氨基酸连接方式的定义(FoldTree/AtomTree)等，这部分构成了蛋白质构象的所有物理信息。(最重要)
Energy: 负责管理氨基酸直接的能量计算所需的信息(EnergyGraph/energies)
ConstraintSet: 负责管理原子间的约束信息(constraints)
DataCache: 负责管理用户自定义的信息

分层式管理使得Pose的信息修改和更新变得容易。

除此以外，还有些外部对象如PDBinfo也负责转换和储存Pose与PDB之间的信息变换。

以下是一个Pose中的架构的示意图：

(图片来源: Leaver-Fay A, Tyka M, Lewis S M, et al. ROSETTA3: an object-oriented software suite for the simulation and design of macromolecules[J]. Methods in enzymology, 2011, 487: 545-574.)

它们之间的关系可以从图上清晰表述: 最底层的是AtomType/ResidueType, 分别定义了原子类型和残基的基本拓扑结构，包括其中的原子成键方式，原子的内坐标等信息。这些信息定义了一个标准的氨基酸残基(Residue)应该包括什么，长什么样子等。氨基酸残基与残基通过树(AtomTree)的方式定义如何与其他残基进行连接，最后得到一个蛋白分子的构象(Conformation)。当然Pose中还有其他的信息如处理能量(Energies)的模块、处理物理约束的模块(ConstraintSet)以及处理中间缓存信息的模块(DataCache)。

三、Pose的读取和生成¶

Rosetta兼容最常规的两种记录结构格式：PDB和Silent文件：

PDB文件可以从https://www.rcsb.org/ 数据库中获取；
Silent文件为Rosetta开发的pose压缩文件（其功能也是储存结构等信息，但其体积比PDB小了10倍之多，非常适合在超算中心进行的数据文件的传输）

Rosetta为PDB的结构信息读入提供了非常丰富的接口，此处我们介绍主要4种结构信息读入相关的函数:

pose_from_pdb：从pdb文件读入并生成pose
pose_from_sequence：从一级序列信息生成pose
poses_from_silent：从silent文件读入并生成pose
pose_from_rcsb：从rcsb数据库远程获取PDB code数据并读入和生成pose

一般而言，Rosetta输入的结构信息大多来源于PDB结构文件，经过处理后生成对应的Pose。以下将逐一介绍如何使用这些外部文件来生成Pose对象。

In [1]:

# 初始化（必须执行的步骤）:
from pyrosetta import init
init()

PyRosetta-4 2021 [Rosetta PyRosetta4.conda.mac.cxx11thread.serialization.python37.Release 2021.31+release.c7009b3115c22daa9efe2805d9d1ebba08426a54 2021-08-07T10:04:12] 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 r292 2021.31+release.c7009b3115c c7009b3115c22daa9efe2805d9d1ebba08426a54 http://www.pyrosetta.org 2021-08-07T10:04:12
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=2087962570 seed_offset=0 real_seed=2087962570 thread_index=0
basic.random.init_random_generator: {0} RandomGenerator:init: Normal mode, seed=2087962570 RG_type=mt19937

1.从PDB文件生成Pose¶

此处以人工设计的小蛋白PDB文件为例读入并生成Pose:

(图片来源: 晶泰科技团队)

In [2]:

from pyrosetta import pose_from_pdb
pdb_pose = pose_from_pdb('./data/pose_demo.pdb')

core.chemical.GlobalResidueTypeSet: {0} Finished initializing fa_standard residue type set.  Created 983 residue types
core.chemical.GlobalResidueTypeSet: {0} Total time to initialize 0.653008 seconds.
core.import_pose.import_pose: {0} File './data/pose_demo.pdb' automatically determined to be of type PDB
core.conformation.Conformation: {0} Found disulfide between residues 2 11
core.conformation.Conformation: {0} current variant for 2 CYS
core.conformation.Conformation: {0} current variant for 11 CYS
core.conformation.Conformation: {0} current variant for 2 CYD
core.conformation.Conformation: {0} current variant for 11 CYD
core.conformation.Conformation: {0} Found disulfide between residues 5 25
core.conformation.Conformation: {0} current variant for 5 CYS
core.conformation.Conformation: {0} current variant for 25 CYS
core.conformation.Conformation: {0} current variant for 5 CYD
core.conformation.Conformation: {0} current variant for 25 CYD

表明，PDB文件自动读入成功，并发现2对二硫键，位于2、11、5、25位。

In [3]:

# 打印Pose的基本信息
print(pdb_pose)

PDB file name: ./data/pose_demo.pdb
Total residues: 26
Sequence: HCFHCRNIRFCSEDEEELRRAREECK
Fold tree:
FOLD_TREE  EDGE 1 26 -1

通过python的print函数，我们可以直接打印pose对象中的一些基本信息，比如这个pose的pdb文件来源是pose_demo.pdb; 其总氨基酸数目是26、氨基酸的序列是HCFHCRNIRFCSEDEEELRRAREECK。并且可以查看这个pose的foldtree连接上下游关系(可先忽略这个概念。)

2.从序列文件生成Pose¶

除了直接从PDB文件中读取所有的原子信息，pyrosetta中也支持直接通过命令行从一级结构信息直接构建线性结构。以下以构建5个连续的丙氨酸结构为例:

In [4]:

# 从一级序列生成线性结构
from pyrosetta import pose_from_sequence
seq_pose = pose_from_sequence('AAAAA')
seq_pose.dump_pdb('./data/5A.pdb')

Out[4]:

True

将pose输出成PDB结构之后，在pymol中打开可查看到, 构建的模型的确是5个丙氨酸的线性结构:

(图片来源: 晶泰科技团队)

3.从PDB代号生成Pose¶

pyrosetta中也有内置相关的函数，可以直接从PDB数据中下载某个PDB编号，并进行对应的清洗，得到Pose的方法。

In [5]:

# 我们还可以从rscb的PDB数据库中直接下载并读入生成Pose
from pyrosetta.toolbox.rcsb import pose_from_rcsb
rcsb_pose = pose_from_rcsb('4R80')

core.import_pose.import_pose: {0} File '4R80.clean.pdb' automatically determined to be of type PDB
core.conformation.Conformation: {0} [ WARNING ] missing heavyatom:  OXT on residue SER:CtermProteinFull 76
core.conformation.Conformation: {0} [ WARNING ] missing heavyatom:  OXT on residue SER:CtermProteinFull 152

执行完代码后，你会发现当前文件夹中出现了4R80.clean.pdb和4R80.pdb两个PDB结构。使用pymol打开查看。会发现执行clean之后的文件中，底物和水分子均被删除，侧链缺失的原子也被Rosetta自动补全。

(图片来源: 晶泰科技团队)

4 从Silent文件读取生成Pose¶

RosettaSilent文件文件是一种Pose的压缩格式文件，可以用它存储大量的Pose信息，它的体积只有PDB文件的1/10。因此非常适合用于云端超算结果的储存和返回。但是Silent文件并不是人类可读的文件格式。需要使用额外的Rosetta app或从pyrosetta中进行读取。

silent文件内容的举例:

SEQUENCE: AAAAA
SCORE:     score description
REMARK BINARY SILENTFILE
SCORE:     0.000       AAAAA
ANNOTATED_SEQUENCE: A[ALA:NtermProteinFull]AAAA[ALA:CtermProteinFull] AAAAA
LAAAAAAAAAAAAAAAAH/pu/AAAAAAAAAAA5DJAA9By1+DAAAAAz9Bo/kW+YAEAAAAAUBn//Q90F9767l5v8iuq+SPUx9r9XUQp8iuq+SPUx7Ta8D1P8iuq+SPUx7Ta8D1vKkf5/QGs677OEn2PNlPRAZShD9Lpxe5vEGQ0/0ntm/rH6O5vG8Q0/QHLd6Lj/dAw AAAAA
LPdUVAxrmE/jWThSJYSzfABiq1AULO60Ig9BsAhhUsAEkVXeJOUPwAV2LK/jEKQpJeHkYARaLqB067l5PhCZeA92Oz8zSuWgJHMPbAJ5GXB0OEn2vs6QgAxiHUCEpxe5P31SHAdnfzBkH6O5PbGrcA12gIBEj/dAQ AAAAA
LrbnxA14q0B0s39TJd7R9A14q0BUOqkjJ0+1AB58xnCkufCKJI7n7Atn0GDE0HBYpAmwABhpNDB067l5v2irtAh8jWCEHYfSJuU//ABYVVB0OEn2PJnHFBNkyDBEpxe5vMdr+A9EUBAkH6O5vtqr+A5fBhBEj/dAw AAAAA
LJ1IGBZHfrCk1ckiJbiwIBZtKVDEjCF+oms0OBJtfQDkXeboJ9X7QBBa4sCkM0TxJtv8GBlJbvD067l5Pe+ZIBRTvQC02lztJ3gnHB944lD0OEn2vML8IBlfOHEEpxe5PTboCBJQF0DkH6O5Pcf+HB13leDEj/dAQ AAAAA
LrbnRB14q0D0s39bJkrcXB14q0DUOqkrJpspZBpsDREkufCSJAL2ZB9VDbEEKAM4PIUMbBVs9YEEKAM4v2TkZBJR8bD067l5vRfpPBdscIEE8CabpMYzYBdIAlD0OEn2P+U7dBhuOcDEpxe5vccJYB5e/6CkH6O5vMjJYBZM2qDEj/dAw AAAAA

In [6]:

# 读入Silent文件
from pyrosetta.io import poses_from_silent
poses = poses_from_silent('./data/demo.silent')
poses = list(poses)
for pose in poses:
    print(pose)

core.io.silent.SilentFileData: {0} Reading all structures from ./data/demo.silent
core.io.silent.SilentFileData: {0} Finished reading 2 structures from ./data/demo.silent
core.conformation.Conformation: {0} Found disulfide between residues 2 11
core.conformation.Conformation: {0} current variant for 2 CYD
core.conformation.Conformation: {0} current variant for 11 CYD
core.conformation.Conformation: {0} current variant for 2 CYD
core.conformation.Conformation: {0} current variant for 11 CYD
core.conformation.Conformation: {0} Found disulfide between residues 5 25
core.conformation.Conformation: {0} current variant for 5 CYD
core.conformation.Conformation: {0} current variant for 25 CYD
core.conformation.Conformation: {0} current variant for 5 CYD
core.conformation.Conformation: {0} current variant for 25 CYD
core.conformation.Conformation: {0} Found disulfide between residues 2 11
core.conformation.Conformation: {0} current variant for 2 CYD
core.conformation.Conformation: {0} current variant for 11 CYD
core.conformation.Conformation: {0} current variant for 2 CYD
core.conformation.Conformation: {0} current variant for 11 CYD
core.conformation.Conformation: {0} Found disulfide between residues 5 25
core.conformation.Conformation: {0} current variant for 5 CYD
core.conformation.Conformation: {0} current variant for 25 CYD
core.conformation.Conformation: {0} current variant for 5 CYD
core.conformation.Conformation: {0} current variant for 25 CYD
PDB file name: /Users/kunkun/Desktop/PyRosetta培训资料/data/pose_demo.pdb
Total residues: 26
Sequence: HCFHCRNIRFCSEDEEELRRAREECK
Fold tree:
FOLD_TREE  EDGE 1 26 -1 
PDB file name: /Users/kunkun/Desktop/PyRosetta培训资料/data/pose_demo1.pdb
Total residues: 26
Sequence: HCFHCRNIRFCSEDEEELRRAREECK
Fold tree:
FOLD_TREE  EDGE 1 26 -1

运行结果表明，我准备的silent文件里面有2个Pose。通过python的list索引或for循环语句，我们可以轻松地访问每一个Pose的信息。

In [7]:

# 索引第二个Pose
first_pose = poses[1]
print(first_pose)

PDB file name: /Users/kunkun/Desktop/PyRosetta培训资料/data/pose_demo1.pdb
Total residues: 26
Sequence: HCFHCRNIRFCSEDEEELRRAREECK
Fold tree:
FOLD_TREE  EDGE 1 26 -1

四、Pose的输出¶

从上面的4个例子中，我们已经学会了如何从Seq,PDB以及Silent文件中读入并生成Rosetta的Pose。那么接下来的最后一个例子就讲述一下如何从Pose输出对应的信息。

从已经生成的pose对象中输出PDB结构或Silent文件，仅需要调用Pose类中dump_pdb的方法函数即可或poses_to_silent函数即可。

In [8]:

# 从Pose输出pdb
pdb_pose.dump_pdb('./data/output.pdb')

Out[8]:

True

In [9]:

# 输出5个重复的pose到silent文件中
from pyrosetta.io import poses_to_silent
for i in range(5):
    poses_to_silent(pdb_pose, './data/multi_AAAAA.silent')

进阶Pose IO的技巧¶

1. PDB结构清洗¶

使用toolbox中的cleanATOM对PDB进行数据清洗，仅把ATOM和TER行进行分离:

In [10]:

from pyrosetta.toolbox import cleanATOM
cleanATOM("./data/4R80.pdb", out_file="./data/4R80_clean2.pdb")

2. 读入带有非标准氨基酸或氨基酸修饰的PDB结构¶

如以PDB:4jfx结构为例，该结构中存在名为PTR(磷酸化酪氨酸)的残基。

Rosetta中内置可以识别的被修饰氨基酸有3种: PTR,SEP,TPO.

In [11]:

# 需要额外读入ptm-caa的参数文件:
init("-extra_res_fa ptm-caa/PTR.params")
ptms_pose = pose_from_pdb('./data/4jfx_peptide.pdb')
print(ptms_pose.sequence())
ptms_pose.dump_pdb('./data/ptm.pdb')

PyRosetta-4 2021 [Rosetta PyRosetta4.conda.mac.cxx11thread.serialization.python37.Release 2021.31+release.c7009b3115c22daa9efe2805d9d1ebba08426a54 2021-08-07T10:04:12] 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 r292 2021.31+release.c7009b3115c c7009b3115c22daa9efe2805d9d1ebba08426a54 http://www.pyrosetta.org 2021-08-07T10:04:12
core.init: {0} command: PyRosetta -extra_res_fa ptm-caa/PTR.params -database /opt/miniconda3/lib/python3.7/site-packages/pyrosetta/database
basic.random.init_random_generator: {0} 'RNG device' seed mode, using '/dev/urandom', seed=-1317805105 seed_offset=0 real_seed=-1317805105 thread_index=0
basic.random.init_random_generator: {0} RandomGenerator:init: Normal mode, seed=-1317805105 RG_type=mt19937
core.import_pose.import_pose: {0} File './data/4jfx_peptide.pdb' automatically determined to be of type PDB
GNYVVTYA

Out[11]:

True

(图片来源: 晶泰科技团队)

3. 读入带有金属离子的PDB文件¶

参考: https://www.rosettacommons.org/docs/latest/rosetta_basics/non_protein_residues/Metals

目前Rosetta能够自动处理的一些常见金属离子: Zn, Cu, Fe, Mg, Na, K, Ca等。但注意的是，rosetta并没有认识金属和氨基酸残基形成化学键，仅是在处理是加了坐标约束。

In [12]:

# 在初始化的时候，需要设置metals的检测。
init("-in:auto_setup_metals")
metal_protein_pose = pose_from_pdb('./data/3cjk.pdb')
metal_protein_pose.dump_pdb('./data/3cjk_metalsetip.pdb')

PyRosetta-4 2021 [Rosetta PyRosetta4.conda.mac.cxx11thread.serialization.python37.Release 2021.31+release.c7009b3115c22daa9efe2805d9d1ebba08426a54 2021-08-07T10:04:12] 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 r292 2021.31+release.c7009b3115c c7009b3115c22daa9efe2805d9d1ebba08426a54 http://www.pyrosetta.org 2021-08-07T10:04:12
core.init: {0} command: PyRosetta -in:auto_setup_metals -database /opt/miniconda3/lib/python3.7/site-packages/pyrosetta/database
basic.random.init_random_generator: {0} 'RNG device' seed mode, using '/dev/urandom', seed=-667901329 seed_offset=0 real_seed=-667901329 thread_index=0
basic.random.init_random_generator: {0} RandomGenerator:init: Normal mode, seed=-667901329 RG_type=mt19937
core.import_pose.import_pose: {0} File './data/3cjk.pdb' automatically determined to be of type PDB
core.io.pose_from_sfr.PoseFromSFRBuilder: {0} CD 171 was added by a jump, with base residue 81
core.conformation.Conformation: {0} [ WARNING ] missing heavyatom:  OXT on residue ILE:CtermProteinFull 68
core.conformation.Conformation: {0} [ WARNING ] missing heavyatom:  OXT on residue ARG:CtermProteinFull 143
core.util.metalloproteins_util: {0} Automatically setting covalent bonds between metal ions and metal-binding residues.
core.util.metalloproteins_util: {0} Adding covalent linkage between residue 11's  SG  atom and residue 144's CD   atom.
core.util.metalloproteins_util: {0} Adding covalent linkage between residue 80's  OG1 atom and residue 144's CD   atom.
core.util.metalloproteins_util: {0} Adding covalent linkage between residue 81's  SG  atom and residue 144's CD   atom.
core.util.metalloproteins_util: {0} Adding covalent linkage between residue 84's  SG  atom and residue 144's CD   atom.
core.util.metalloproteins_util: {0} Automatically setting up constraints between metal ions and metal-binding residues.

Out[12]:

True

(图片来源: 晶泰科技团队)

在接下来的章节，我们会介绍更多关于Pose的不同Layer的工作方式以及其中的信息提取方法。

In [ ]: