永远相信美好的事情终将发生

surflex对接基础

发表于 分类于

surflex 2.7 包括surflex-dock 和 surflex-sim. The former combines a refined descendant of the Hammerhead scoring function coupled with the alignment/conformation optimization proce-dures implemented for morphological similarity. The latter addresses aspects of computational drug design in the absence of a protein structure.

Notes:
使用surflex-dock 探测蛋白表面的沟槽？？
使用surflex-sim 计算分子相似性？？
对其它的结合态结构进行打分
将分子拆分为fragment
align复合物
提取口袋和配体

在sybyl-x-2.1 中包含v2.706.13294 版本的surflex，最新版本为Surflex v3.066，后续的surflex还增加了 surflex-qmod模块，具体见本文最下方说明。

surflex-dock 入门

Notes:

  • 对于虚拟筛选通常使用 “-pscreen” 或者 “-lscreen” 选项，对于最优构象预测通常使用 “-pgeom”选项，对于以上两种情况使用 ring search (“+ring”) 均能够提升计算结果(同时会消耗更多时间).
  • 基本的参数为 -pscreen 和 -pgeom. 前者多用于虚拟筛选，自动添加参数 +premin 和 +remin，二者分别turn on ligand minimization prior to docking and all-atom in-pocket minimization after docking.
    -pgeom 则包括 -pscreen的内容，并添加了 -multistart 4, -ndock_final 20 和 -div_rms 0.5 三个参数。
  • 建议使用Sybyl mol2 结构作为输入，同时也可使用MDL或者sd格式(小分子)，PDB(蛋白，容易出错)。所有输入文件需要加氢(包括非极性氢)
  • surflex-dock 允许 multi-structure docking (multiple protein conformations for a single target), and protein pocket flexibility.

流程:

  • 构建protomol
  • 对接
  • 结果分析及后处理

proto command: Protomol Generation

形成protomol的两种方式: ligand-based 和 residue-based

1
2
3
surflex-dock.exe proto ligand_reference.mol2 protein.mol2 p1  # ligand-based

surflex-dock.exe resproto residue-list protein.mol2 p1        # residue-based

如果使用基于残基的方式，需要指定 residue-list 文件，格式如下:
1
2
3
4
5
ILE1
VAL2
ILE118
GLY122
ASN123

输出文件均以p1为前缀，包括 protomol(p1-protomol.mol2), 单个的 protomol fragment(p1-probe.mol), 用来定义protomol空间的信息文件(p1-marked.pdb),用来观察参数的信息文件(p1-thresh.mol)

Notes: When using a ligand to specify an active site, by default, the voxels occupied by the ligand are explored by the protomol even if they are not highly buried. This can be turned off by employing the -mark_lig switch.

如果既无配体又不确定活性位点，可以使用如下命令

1
surflex-dock.exe proto none protein.mol2 p1

none选项会让surflex探索蛋白表面的口袋，并在每一个口袋处形成对应的component文件(p1-comp-*.mol2)，用于形成protomol:

1
surflex-dock.exe proto p1-comp-002.mol2 protein.mol2 p2

两个参数可用来控制protomol的范围：
proto_thresh (default 0.5, 0-1): determines the degree of buried-ness for the primary volume used to generate the protomol. 值越大则体积越小。文件p1-marked-thresh-<n>.pdb对应于不同临界值的空间。

proto_bloat (default 0): indicates how far beyond the primary volume (in Angstroms) the protomol volume should be expanded。

所以，使用如下命令可得到more extended protomol

1
surflex-dock.exe -proto_thresh 0.2 -proto_bloat 2 proto ligand_reference.mol2 protein.mol2 p1

Smaller protomols yield faster searches, and it is not the case that the docked ligands are strictly limited to the volume of the protomol.

另外一个影响protomol范围的参数adaptive protomol switch:

1
surflex-dock.exe +adapt_proto –adapt_thresh 0.4 proto lig.mol protein.mol2 p1

With this procedure, Surflex-Dock will build a protomol that gives a fixed degree of coverage against the residues that are proximal to the ligand (as specified) or explicitly listed residues (using the resproto command). Higher values of adapt_thresh yield higher degrees of coverage.

dock command: Docking a Single Molecule

1
2
dock ligand protomol protein
surflex-dock.exe dock ligand.mol2 p1-protomol.mol2 protein.mol2

会输出打分值最高的十个构象，按打分值降序排列(final-.mol2)。每个构象有两个打分值 affinity (-log(Kd)) 和 crash score (also pKd units). Crash score接近于0更好，*报告的分数已经包括 crash score。

三个参数能够明显改善对接质量:

  1. -multistart <n> 以n个初始位置作为起点进行对接，然后返回最佳结果。
    For flexible molecules, since the search is not exhaustive, using multiple starting points will frequently yield higher-scoring and more consistent results, independent of initial starting pose. Generally speaking, -multistart 10 is as high as one sees returns on the investment of time, with the plateau beginning at –multistart 4.
  2. -ndock_final <n> 确定最终输出的top n个构象。
  3. The third option affects the density of alignment search. The new method can be controlled in terms of search density with the –spindense parameter (higher numbers indicate more dense search) as well as the -nspin parameter (higher numbers indicate denser sampling of axial rotations).

使用 +pflex 选项打开蛋白口袋的柔性，允许氢原子来adapt。默认的covalent force-field 强度为0.1(值比较低，允许氢原子的显著移动来适应配体结合)，可以使用 +hprot 选项来允许结合口袋中所有原子的移动，包括重原子，此时建议提高protein covalent force-field的值(eg. -pcov 0.6)来约束原子的移动。

opt command: 优化已经对接的构象

1
2
opt ligand protein
surflex-dock.exe opt ligand.mol2 protein.mol2

输出的优化构象为 opt.mol2.

dock_list command: Docking a Molecule List

1
2
dock_list liglist protomol protein log
surflex-dock.exe dock_list ligarchive p1-protomol.mol2 protein.mol2 log

若ligarchive具有.mol2扩展名，则认为它是包含多个分子的mol2文件
若ligarchive具有.sd扩展名，则认为他是包含多个分子的MDL sd 格式文件
否则，认为其内容为分子路径，每行一个分子

输出的log文件中，三个打分值(affinity, crash, and polar contribution<氢键相关>)，对应的构象输出为log-results.mol2

-fmatch option: Docking using Placed Fragments

surflex允许基于分子片段的位置进行对接，通过将目标分子ligand.mol2中的片段与给定的frag.mol2中片段进行align来实现

1
2
-fmatch fmol
surflex-dock.exe -fmatch frag.mol2 dock ligand.mol2 p1-protomol.mol2 protein.mol2

默认设置中，忽略frag.mol2中的氢原子的align， +fhmatch选项强制加上氢原子的align。
frag.mol2中允许存在多个片段。
在对接多个分子时，默认设置中，不含有对应片段的目标分子将按照正常模式进行对接。 +fskip 选项可以覆盖这种行为，从而自动跳过不含对应片段的分子。
-cpen 选项确定片段align时的约束力(eg. -cpne 100)。
若同时提供protomol和frag.mol2，默认将同时使用二者。如果希望只使用fragment，不使用protomol，可使用 -fdockreg 选项。
The fragment options allow for coarser conformational search (+fcoarse), varying the depth of initial conformational search from the placed fragment (-fidepth [default 3]), and varying the depth of successive conformational enumerations (-frdepth [default 3]).

fragmentize command: 将已知的配体分子拆分为fragments

使用 fragmentize 命令，将已知的配体分子拆分为fragments， +misc_ring 选项也会将环拆开作为片段
使用 choose_frags 命令，由多个fragment中选择一个更小系列的fragments，作为 -fmatch的输入。

1
2
3
4
5
6
7
8
9
10
11
12
fragmentize molarchive outprefix
choose_frags fragarchive aligned_mols outprefix

surflex-dock.exe fragmentize pde4b/train-ligands/ligand-1.mol2 pde4b/train-ligands/frag-1
surflex-dock.exe fragmentize pde4b/train-ligands/ligand-2.mol2 pde4b/train-ligands/frag-2
surflex-dock.exe fragmentize pde4b/train-ligands/ligand-3.mol2 pde4b/train-ligands/frag-3
surflex-dock.exe fragmentize pde4b/train-ligands/ligand-4.mol2 pde4b/train-ligands/frag-4
surflex-dock.exe fragmentize pde4b/train-ligands/ligand-5.mol2 pde4b/train-ligands/frag-5

cat pde4b/train-ligands/frag-*.mol2 > pde4b/train-ligands/fragall.mol2

surflex-dock.exe choose_frags pde4b/train-ligands/fragall.mol2 pde4b/train-ligands/ligandall.mol2 pde4b/train-ligands/chosenfrag

mdock_list command: Docking to Multiple Protein Conformations

1
2
mdock_list liglist targpath log
surflex-dock.exe mdock_list ligarchive Targets log

Targets 文件描述蛋白的信息及对应的protomol，允许每个蛋白构象使用单独的protomol，文件格式如下:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
Nproteins 5
protein-opt-1.mol2 Nprotomol 2
p-opt-1-protomol.mol2
train-ligands/ligand-opt-1.mol2
protein-opt-2.mol2 Nprotomol 2
p-opt-2-protomol.mol2
train-ligands/ligand-opt-2.mol2
protein-opt-3.mol2 Nprotomol 2
p-opt-3-protomol.mol2
train-ligands/ligand-opt-3.mol2
protein-opt-4.mol2 Nprotomol 2
p-opt-4-protomol.mol2
train-ligands/ligand-opt-4.mol2
protein-opt-5.mol2 Nprotomol 2
p-opt-5-protomol.mol2
train-ligands/ligand-opt-5.mol2

表明具有5个受体蛋白，对应的结构文件为protein-opt-*.mol2。
每个蛋白指定两个protomol，一个为标准的protomol，一个为cognate ligand 直接作为protomol。
1
2
3
surflex-dock.exe mdock_list test-ligand/testlig.mol2 Targets logdef

surflex-dock.exe -fmatch train-ligands/frag.mol2 mdock_list test-ligand/testlig.mol2 Targets logdef

Self Scoring

对于比较大的配体，其分子内的非共价相互作用可能对对接的打分值有影响。故可以使用 +self_score 选项来引入这种自身相互作用的影响。
当使用 +pflex 选项的时候会，会对蛋白自动打开 self score，来计算蛋白构象的adaptation。

Protein Pocket Alignment

基于表面的蛋白结合口袋align，四个命令: psim_matrix, psim_buildtree, psim_one, and psim_list

1
surflex-dock.exe psim_matrix ProteinList palign

ProteinList 文件每一行包含蛋白和对应的配体的路径，配体用来定义蛋白的口袋。输出文件为 palign-results，包含align的信息。接下来构建 optimal tree of alignments
1
surflex-dock.exe psim_buildtree palign-results out

输出文件为 out-lig-.mol2 和 out-.mol2 包含align之后的配体和蛋白结构。
out-ligands-aligned.mol2 和 out-proteins-aligned.mol2 包含align之后的结构(蛋白只有结合口袋)，多个结构在一个文件中。

psim_one and psim_list offer the opportunity to align one protein to another or a list of proteins to a single one and function analogously to above, without the requirement to separately build a final alignment.

Note also that the psim_buildtree command produces a file suitable for generating a visual depiction of the alignment tree using the Dot program from the GraphViz collection (out-tree.dot in the example above).

1
dot -Tpdf -o psimlog-tree.pdf out-tree.dot

Protein Pocket Preparation

Frequently, a protein structure will exhibit significant clashes with a cognate ligand. 所以可以通过pprep_protons 或pprep_all命令来调整结合口袋中氢原子或所有原子的位置来消除clashes。

1
surflex-dock pprep_protons ligand protein outprefix

输出文件为outprefix-ligand.mol2, outprefix-protein.mol2, outprefix-protein-trim.mol2(只含有口袋)。

Scoring Function Optimization

1
optimize logprefix constraint-file init-params

Surflex-Dock Version 2.4 and higher offers the opportunity to tune the Surflex-Dock scoring function based on additional data. The data can be positive, which includes protein/ligand complexes with known affinity. The data can also include negative information.

Ring Flexibility

Surflex-Dock implements in-lined ring flexibility in a general way. The +ring option turns this procedure on. The behavior is modified by the -rthresh parameter (kcals above global energy minimum beyond which ring conformations are not kept).
Turning on ring search (“+ring”) can yield improved results for both screening and pose predic-tion at the expense of some additional computational cost.

Post-Processing Results

logprocess command

通常，我们基于overall score来对对接结果进行排序，但也可以通过以下两种方式进行排序:
1) a derived combination score, and 2) a method for placing thresholds on the crash and polar score

比如，对 polar score添加1.0的临界值，对crash score添加-1.0的临界值，对可旋转键的数目添加100的临界值(通常，高柔性的配体更容易是假阳性)

1
2
logprocess logfile
surflex-dock.exe -lp_polar 1.0 -lp_crash -1.0 -lp_rot 100 logprocess log

输出文件为 log-processed

A combination score is given first, which combines the reported score and crash values.
The threshold that is supplied on the command line for crash (-1.0 in the example above) is allowed “for free” so the amount of crash that is below that level is given back to the affinity score. So, with an observed crash of –2.0, a thresh-old of –1.0, and an affinity score of 7.1, the combination score would be 8.1.
The smaller the crash threshold, generally the better able Surflex is to reject false positives. However, this may come at the expense of some true positives that have particularly tight fits into the protein active site in question.

With Version 1.31 and later, we recommend running the logprocess command with all default parameters. This yields no change from the reported score in the log file and has been used for the results in all protein screening enrichment benchmark tests.

logprocsdf command

使用 logprocsdf 选项将log文件和相应的mol2文件转换为sdf文件

1
2
logprocsdf logfile
surflex-dock.exe logprocsdf log

get command 获取特定的分子

1
2
get mol2archive molname outmolname
surflex-dock.exe get log-results.mol2 log-0 foo.mol2

mget command

mget 选项具有相同的格式，但第二个参数是一个一个文件，内容为待提取的分子名称。

1
mget mol2archive molnamelist outmolarchive

score_list command

处理其它对接程序的对接结果，或者是手动放置的分子

1
2
score_list liglist protomol protein log
surflex-dock score_list ligarchive.mol2 protomol.mol2 protein.mol2 log

先进行打分，然后对构象进行优化，之后再进行打分，并输出log文件和对应的优化之后的构象。

rescore_run

将会基于提供的log文件对给定的对接进行重打分。
用于检测不同的打分设置 (e.g. with –lparam) 或者打开蛋白柔性 (e.g. +pflex ，在某些特定的活性位点中对结果影响比较大)对打分的影响。

1
2
rescore_run logfile protomol protein prefix
surflex-dock.exe +pflex +hprot +pcov 0.6 rescore_run log protomol protein rescoreheavy

rescore_multi

适用于multi-protein docking run

1
2
rescore_multi logfile targpath prefix
surflex-dock.exe +pflex +hprot +pcov 0.6 rescore_multi log Targets rescoreheavy

Other commands:

prot 对分子进行质子化

1
2
prot mol_or_mollist output_prefix 
surflex-dock.exe +misc_remin +fp +misc_ring -misc_outconfs 1 prot inputarchive.mol2 outputarchive

It is suggested to make use of +misc_remin in order to eliminate conformational bias that may affect different scaffolds differently and lead to bias in results.
Note: +misc_ring will generate ring conformations as well. If –misc_outconfs is set to greater than 1, then each molecule will generate an individual *-ring.mol2 file in addition to the single conformation in the molecule archive.
-fp 选项设定删去所有的氢原子，然后加氢。

min

1
min mol_or_mollist output_prefix

功能与 prot 相似，但不进行加氢操作，故可以使用其它程序先加氢。

reorder

输入一系列分子，然后根据柔性对其进行排序(从柔到刚)

1
reorder mol2archive proportion outputarchive

search 搜索 acyclic bonds

1
search mol1 mol2...

random 随机摆放小分子

1
2
3
random mol1 mol2...

-misc_outconfs <n>   # 产生n个随机构象

rms

RMS computes rmsd between mol1 and mol2

1
rms mol1 mol2

rms_list

1
rms_list multi_mol reference_mol logfile (appends information on rms of confs to reference)

trim

1
trim protein ligand outprotein distance

info 提供分子的信息

1
info mol

posefam 对对接的构象进行分类

1
surflex-dock.exe posefam logfile

Other options:

-multiproc # 指定在n个处理器上运行

-lparam # This parameter takes a file as input that contains alternate scoring function parameters for Surflex-Dock. The file “default.param” contains the default pa-rameters for Surflex-Dock, and it is at the top-level in the software distribution.

Surflex-Sim:

基于配体的相似性筛选
步骤:

  1. 基于多个配体形成 ligand-based hypothesis
  2. 基于 hypothesis 利用相似性来筛选分子
  3. 结果后处理

Pre-Searching Molecules

search_library 命令，一次性将多个分子结构转化为library文件，后续可被-lscreen和-lscreenopt使用。

1
surflex-sim.exe search_library mol_list pre

mol_list 文件中每一行分别为输入分子的路径，比如:
1
2
3
4
5
BasisMols/basis01.mol
BasisMols/basis02.mol
BasisMols/basis03.mol
BasisMols/basis04.mol
BasisMols/basis05.mol

输出文件为pre-.mol2，单个mol2中包含多个构象(最大数量由 -sl_nconfs 指定，默认为200)；pre-.mol2_im，内容为20维的molecular imprint。使用默认参数即可。

Aligning Molecules

1
surflex-sim.exe align mol1.mol mol2.mol

输出的 aligned 构象文件为final-.mol2，其对应的打分值高于0.7通常能够表明分子之间的功能关系。

多个分子的相互比对

1
surflex-sim.exe hypo liglist log

若liglist具有.mol2扩展名，则认为它是包含多个分子的mol2文件
若liglist具有.sd扩展名，则认为他是包含多个分子的MDL sd 格式文件
否则，认为其内容为分子路径，每行一个分子

如果输入分子具有柔性环，还能使用 +ring 参数。
输出信息还包括 scores and internal clash values

Generally, the first hypothesis (hypo-0) is the most sensible to use. However, when something is known about the SAR of analogs of the ligands used to generate a hypothesis, a better selection of hypothesis may be possible by browsing the top scoring ones

Lists of molecules can be aligned to either a single molecule or to mutually aligned sets of molecules using align_list:

1
2
3
surflex-sim align_list LigList1 TargList log

surflex-sim +ring align_list TestAll.mol2 BestHypo.mol2 logtest

-nsim_final 参数控制输出的最终构象的个数。

The –fscreen option yields very fast similarity computations. It does this by turning off pre-minimization, all-atom post alignment optimization, eliminating molecular fragmenta-tion in favor of conformer sampling, and by limiting the degree of local pose optimization performed in the last stages of molecular alignment. This can be an especially useful parameter-ization for going through large databases quickly, especially if followed by a more thorough screen of the top fraction of ligands using more thorough settings. The –pscreen option is the recommended setting for screening, and it is fast enough for large databases in situations where multiple processors are available. For detailed studies of relative alignments, the –pgeom option is suggested, and where ring geometries are also to be considered the –pgeomx option is preferred.

Reference Sets and Similarity Vectors

比对和构象优化的速度都比较快，但是大规模的相似性计算比较耗时，这里引入一种vector-based 近似，将每个分子表示为相对于一个 reference set of molecules的 相似性向量，进而计算分子之间的形态相似性。

1
surflex-sim.exe vector liglist reflist vectorfile

liglist 包含待比较的分子(内容为小分子的存储路径)
Reflist is a well-chosen set of (usually 20) molecules in fixed conformations and alignments.
输出的vectorfile中每一行为一个lig分子相对于每一个reference 分子的相似性
然后通过计算两个向量之间的Euclidean distance，即可获得两个分子的相似性。

-multiproc 指定使用多核，-fscreen 能提升速度和精确性

选择合适的reference set比较重要

1) the molecules must be orthogonal (i.e. dissimilar from each other); and 2) the molecules must come from the population of molecules for which comparisons are to be made

In the case of small-molecule drugs, selecting a diverse reference set from the CMC database (available from MDL) has worked well.

The basis set of molecules used in Cleves and Jain (2006) is included in the distribution (see the BasisMols folder under the Similarity examples). Note that we do not suggest that this set will be generally useful for all populations of molecules. We have found that the set works well for reproducing similarities within the space of approved therapeutics. It will probably be the case that for specific collections, different sets may be better.

如果你想在一个liglist中选出reference set，使用如下命令可以从中选出N个最有可能的orthogonal molecules:

1
surflex-sim choose_ref liglist reflist N

比较耗时，一个有效的方法是将liglist随机划分为多个sub-list，每个sub-list中分子数目为最终reference set的3-5倍。

Molecular Similarity Examples

1
2
3
4
5
6
7
8
9
surflex-sim.exe align mols/ligand1tng.mol2 mols/ligand3ptb.mol2

surflex-sim hypo LigList3 log

surflex-sim align_list LigList1 log-hypo0.mol2 log 

surflex-sim choose_ref RefStart RefList 10 

surflex-sim vector LigList2 RefList LigVector

SURFLE-XQMOD TECHNICAL MANUAL

The SurflexQMOD set of algorithms integrates underlying ideas and algorithms from molecular similarity [1–6], molecular docking [7–14], and multipleinstance learning [7, 11, 13, 15–18] in order permit the construction of protein binding site analogs. The theory and use of the method for binding affinity prediction and iterative lead optimization is discussed in the companion book to this manual as well as several papers [19–22].

surflex-dock 详细参数列表:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
Usage: surflex-dock <options> <command> args
  PROTOMOL AND PROTEIN PREPARATION COMMANDS:
    proto sitemol protein protomol
    resproto reslist protein protomol
      -proto_thresh (0.5) Threshold for marking a voxel based on buriedness
      -proto_bloat (0) Number of voxels to bloat protomol
      +mark_lig (+-) Specifies ligand voxel marking
      -adapt_proto (+-) Specifies adaptive protomol thresholding
        -adapt_thresh (0.2) Modifies adaptive protomol threshold
      -lparam file Loads a docking parameter file output from Surflex-Dock-Optimize v1.0
      pprep_all ligand protein outprefix
      pprep_protons ligand protein outprefix
        -nrepeat (5) Number of iterations
    
  DOCKING and SCORING COMMANDS:
    dock ligand protomol protein
    mdock_list liglist targpath log
    dock_list liglist protomol protein log
      -pscreen Screening parameter set (+premin +remin -ndock_final 3).
      -pgeom Docking accuracy parameter set
         (-spindense 6.0 +premin +remin -multistart 4 -div_rms 0.5 -ndock_final 20).
      -pgeomx Turns on more exhaustive docking accuracy parameter set
        (-spindense 9.0 +premin +remin -multistart 6 -div_rms 0.5 -ndock_final 20 +ring).
      -self_score (+-) Specifies self-scoring within the scoring function.
      +spinalign (+-) Specifies spin alignment method).
        -spindense (3.0) Density of search for spinalign.
        -nspin (12) Number of spins per alignment.
      -fmatch fmol Turns on substructural fragment docking
        -fskip (+-) Specifies skipping of mols that have no substructure matches)
        -fhmatch (+-) Specifies matching hydrogens in fragment
        +fdockreg (+-) Specifies to dock using the protomol as well as fragments
        -fidepth (3) Rotatable bond depth for initial fragment
        -frdepth (3) Rotatable bond depth for recursive fragment search
        -fcoarse (+-) Specifies coarse conformational search in fmatch
        -cpen (0.0) Penalty for deviating from fragment(non-negative)
      -ndock_final (20) Final number of docking poses for output
      -ring (+-) Specifies dynamic ring search)
        -rthresh (5.0) Threshold for ring energy delta (kcal)
      +premin (+-) Specifies preminimization
      +remin (+-) Specifies reminimization
      -multistart (0) Number of additional starting confs per mol
      -maxrot (100) Max number of rotatable bonds on which to operate
      -div_rms (0.05) RMS minimum difference of final poses
      -bfgs_focus (1.00) Thoroughness of local optimization (bigger is more thorough)
      +frag (+-) Specifies ligand fragmentation
      -rigid (+-) Specifies treatment of ligands as rigid
      -protodthresh (3.0) Max distance from protomol observer to protein
      -lparam file Loads a docking parameter file output from Surflex-Dock-Optimize v1.0
      +soft_box (+-) Specifies soft grid treatment
      -multiproc npc pnum Indicates NPC processor run, current processor is PNUM
      +fastsearch (+-) Specifies fast conformer search method
        -maxdiff (20) Max number of different confs to yield per fragment
        -maxsamp (300) Max number of sampling confs per recurrence
        -maxrecur (10) Max depth of recurrence
      -maxconfs (20) Max number of confs per fragment when using -fastsearch
      -grid_bloat (6) Size in Angstroms to bloat the grid
    score_list liglist protomol protein log
      -lparam file Loads a docking parameter file output from Surflex-Dock-Optimize v1.0
      -misc_premin (+-) Specifies score_list preminimization
    rescore_run logfile protomol protein prefix
      -lparam file Loads a docking parameter file output from Surflex-Dock-Optimize v1.0
      -pflex (+-) Specifies protein flexibility in rescore_run, mdock_list, and rescore_multi
    rescore_multi logfile targpath prefix
      -dumpprot (+-) Dump protein conformations for each pose in rescore_multi
      -ntweak (0) Number of times to tweak the ligand for sampling in rescore_multi
      -tweaksize (0.01) Amount of atom position tweak
      -rescorefast (+-) Specifies faster rescoring using less thorough protein flex
      -maxposes (10) Max number of poses to rescore for each ligand
    posefam logfile
      -rmsbin (1.50) RMSD threshold for binning pose families
      -famdiff (0.05) Tanimoto similarity above which pose families are considered redundant
      -poseprob (0.000001) Individual pose probability threshold
      -scorechoice (0) Use: 0 = cscale, 1 = score, 2 = full_ligand, 3 = full+cscale 
    opt ligand protein
      -lparam file Loads a docking parameter file output from Surflex-Dock-Optimize v1.0
      -misc_premin (+-) Specifies opt preminimization
      -misc_remin (+-) Specifies opt reminimization
      -lcov (1.00) Weight of ligand covalent force-field
  
  SCORING FUNCTION OPTIMIZATION AND OPTIONS:
    optimize logprefix constraint-file init-params
    [ Search Strategy: Default is multiple strategies ]
    -lineopt only systematic parameter pertubation
    -randwalk only random walk
    -randsearch only random search
    -stepsize N lineopt param incr = 1/N * initial param value
    -rw_stepsize N randwalk param incr = 1/N * initial param value
    -paramrange N randsearch sampling range = N * initial param value
    
    [ Stopping Conditions ]
    -maxepochs N maximum optimization epochs
    -rw_maxepochs N maximum random walk epochs
    -minmse N minimum MSE goal
    -ncycle N maximum number of parameter search cycles
    
    [ Posecache ]
    -ngood2optpose N optimize pose cache after N good steps
    -posecachesize N number of cached poses kept during optimization
    
    [ Other ]
    -score toggles on score data once and exit
    -exportposes toggles on export of optimized poses
    -repeat N repeat the optimization N times
    -useparam log take first parameter steps from param.log
  
  POST-PROCESSING COMMANDS AND OPTIONS:
    logprocess logfile
      -lp_polar 0.0 (polar score thresh below which scores --> -1000)
      -lp_nrot 100 (rotatable bond thresh above which scores --> -1000)
      -lp_crash 0.0 (non-positive crash value to give back to score)
    logprocsdf logfile (merges log file information with mol2 --> sdf)
    
  FRAGMENT GENERATION:
    fragmentize molarchive outprefix
      -misc_ring (+-) Specifies ring favoritism
    choose_frags fragarchive aligned_mols outprefix
    
  PROTEIN ALIGNMENT:
    psim_matrix ProteinList outprefix
    psim_one p1 l1 p2 l2 outprefix
    psim_list ProteinList p2 l2 outprefix
      -psim_trim (15.0) Distance to residues to trim protein.
    psim_buildtree ResultsFilePath outprefix
    
  MISC-PROCESSING COMMANDS and SPECIFIC ASSOCIATED OPTIONS:
    prot mol_or_mollist output_prefix
      -fp (+-) Specifies full reprotonation
      -misc_remin (+-) Specifies minimization after protonation
      -misc_ring (+-) Specifies ring search for output
      -misc_outconfs (1) Max number of confs
    min mol_or_mollist output_prefix
      -misc_ring (+-) Specifies ring search for output
      -misc_outconfs (1) Max number of confs
    reorder mol2archive proportion outputarchive
    search mol1 mol2...
      -misc_outconfs (1) Number of confs
    random mol1 mol2...
      -misc_outconfs (1) Number of confs
    rms mol1 mol2
    rms_list multi_mol reference_mol logfile (appends information on rms of confs to reference)
    get mol2archive molname outmolname
    mget mol2archive molnamelist outmolarchive
    info mol
    trim protein ligand outprotein distance

surflex-sim 详细参数列表

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
Usage: surflex-sim <options> <command> args
  HYPOTHESIS PREPARATION COMMANDS AND SPECIFIC OPTIONS:
    hypo liglist log (NOTE: See options for align_list to control search)
      -hyposamp (100) number of initial confs per mol in hypo
      [NOTE: Also see options for align_list to control search]
    hvol mol1 mol2...
    opthvol mol1 mol2...
  
  SIMILARITY COMMANDS:
    align mol1 mol2
    align_list liglist targetlist log
      -fscreen Fast screening parameter set (-premin -remin -nsim_final 3 -frag -maxdiff 50, limited final polishing).
    -pscreen Screening parameter set (+premin -nsim_final 3).
    -pgeom Pose accuracy parameter set (-spindense 6.0 +premin +remin -div_rms 0.25 -nsim_final 20).
    -pgeomx Turns on more exhaustive pose accuracy parameter set (-spindense 9.0 +premin +remin -multistart 3 -div_rms 0.25 -ndock_final 20 +ring).
  vector liglist reflist vectorfile
  sim_matrix liglist1 liglist2 log
    +spinalign (+-) Specifies new spin alignment method).
      -spindense (3.0) Density of search for spinalign.
      -nspin (12) Number of spins per alignment.
    -fmatch fmol Turns on substructural fragment docking
      -fskip (+-) Specifies skipping of mols that have no substructure matches
      -fhmatch (+-) Specifies matching hydrogens in fragment
      -fidepth (3) Rotatable bond depth for initial fragment
      -frdepth (3) Rotatable bond depth for recursive fragment search
      -fcoarse (+-) Specifies coarse conformational search in fmatch
      -cpen (0.0) Penalty for deviating from fragment(non-negative)
    -nsim_final (20) Final number of aligned poses for output
    -ring (+-) Specifies dynamic ring search
      -rthresh (5.0) Threshold for ring energy delta (kcal)
    -premin (+-) Specifies preminimization
    -remin (+-) Specifies reminimization
    -multistart (0) Number of additional starting confs per mol
    -maxrot (100) Max number of rotatable bonds on which to operate
    -div_rms (0.05) RMS minimum difference of final poses
    -bfgs_focus (1.00) Thoroughness of local optimization (bigger is more thorough)
    +frag (+-) Specifies ligand fragmentation
    -rigid (+-) Specifies treatment of ligands as rigid
    -multiproc npc pnum Indicates NPC processor run, current processor is PNUM
    +fastsearch (+-) Specifies fast conformer search method
      -maxdiff (20) Max number of different confs to yield per fragment
      -maxsamp (300) Max number of sampling confs per recurrence
      -maxrecur (10) Max depth of recurrence
    -maxconfs (20) Max number of confs per fragment when using -fastsearch
    -sim_polar_strength (1.0) Relative strength of polar component of similarity function (0 to 1)
    
  score_list liglist targetlist log
  choose_ref liglist refout nmols
  sim mol1 mol2
  opt mol1 mol2
  gsim mol1 mol2
  gsim_list liglist targetlist log
    
  LOG-ODDS COMMANDS: [Note that similarity parameters apply (-lscreenopt recommended)]
    logodds list1 list2 log (p-values estimated from a single global distribution)
    simcover list1 list2 log
      -lo_frag (+-) Specifies ligand fragmentation for log odds computation
      -lo_align (+-) Specifies whether to dump alignments from log odds computation
    gsimlogodds list1 list2 normlist log
    gsimcover list1 list2 log
    
  MISC-PROCESSING COMMANDS and SPECIFIC ASSOCIATED OPTIONS:
    search_library list outprefix
      +imprint (+-) Specifies imprint generation after searching
      +sl_premin (+-) Specifies minimization before search
      +sl_remin (+-) Specifies minimization after search
      +sl_ring (+-) Specifies ring search
      -sl_nconfs (200) Specifies number of output confs  (use -maxsamp and -maxrecur to control search)
    
    prot mol_or_mollist output_prefix
      -fp (+-) Specifies full reprotonation
      -misc_remin (+-) Specifies minimization after protonation
      -misc_ring (+-) Specifies ring search for output
      -misc_outconfs (1) Max number of confs
    min mol_or_mollist output_prefix
      -misc_ring (+-) Specifies ring search for output
      -misc_outconfs (1) Max number of confs
    reorder mol2archive proportion outputarchive
    random mol1 mol2...
      -misc_outconfs (1) Number of confs
    rms mol1 mol2
    get mol2archive molname outmolname
    mget mol2archive molnamelist outmolarchive
    info mol