Abstract
Structure-based lead optimization approaches are increasingly playing a role in the drug-discovery process.
Virtual screening by molecular docking has become a largely used approach to lead discovery in the pharmaceutical
industry when a high-resolution structure of the biological target of interest is available. The performance of
two docking programs (Arguslab and Surflex), for virtual database screening, is studied. Surflex is well recognized
commercial package while Arguslab is distributed freely for Windows platforms by Planaria Software.
Comparisons of these docking programs and scoring functions using a large and diverse data set of pharmaceutically
interesting targets and active compounds are carried out. We focus on the problem of docking and scoring
flexible compounds which are sterically capable of docking into a rigid conformation of the receptor. The three
dimensional structures of a carefully chosen set of 300 pharmaceutically relevant protein-ligand complexes were
used for the comparative study. The results show that Surflex outperforms largely Arguslab in all tests studied.
Keywords
docking programs; drug discovery; biological target; comparative study; ArgusLab; Surflex
Introduction
The development and implementation of a range of molecular
docking algorithms, based on different search methods
(Taylor et al. 2002, Halperin et al. 2002) was observed
in the last few years. This approach has had several recent
successes in drug discovery (Sechi et al., 2005; Liu et al.,
2005).
In the field of molecular modeling, docking is a method
which predicts the preferred orientation of one molecule to
a second when bound to each other to form a stable complex
(Lengauer and Rarey, 1996). Knowledge of the preferred
orientation in turn may be used to predict the strength
of association or binding affinity between two molecules
using for example scoring functions.
Docking is frequently used to predict the binding orientation
of small molecule drug candidates to their protein targets
in order to in turn predict the affinity and activity of the
small molecule. Hence docking plays an important role in
the rational design of drugs (Kitchen et al., 2004). Given the
biological and pharmaceutical significance of molecular
docking, considerable efforts have been directed towards
improving the methods used to predict docking.
Evaluation of existing docking algorithms can assist in the
choice of the must suitable docking programs for any particular
study. Effectively, several studies estimating and comparing
the accuracies of protein-ligand programs like Dock,
ICM, Gold have been reported (Perola, E. et al. , 2004 ;Bursulaya, B. D. et al., 2003).
The goal of this study was to evaluate the ability of
ArgusLab, a freely distributed molecular modeling package
in which molecular docking is implemented, to reproduce
crystallographic binding orientations and to compare its accuracy
with that of a widely well established docking package,
Surflex.
Methods
ArgusLab4.0 has fast become a favorite introductory
molecular modeling package with academics mainly because
of its user-friendly interface and intuitive calculation menus
(Thompson, 2004). The ArgusDock docking engine, implemented
in ArgusLab, approximates an exhaustive search
method. Flexible ligand docking is possible with ArgusLab,
where the ligand is described as a torsion tree and grids are
constructed that overlay the binding site. Ligand’s root node
(group of bonded atoms that do not have rotatable bonds) is
placed on a search point in the binding site and a set of
diverse and energetically favorable rotations is created. For
each rotation, torsions in breadth-first order are constructed
and those poses that survive the torsion search are scored.
The N-lowest energy poses are retained and the final set of
poses undergoes coarse minimization, re-clustering and ranking.
Surflex1.3 method is based on an incremental construction.
There are three steps in performing dock : choosing
how to identify the active site of the protein and constructing
a docking target to which to match molecules (called a
protomol), docking one or many molecules and post-processing
the results.
Each of the basic tasks is controlled by a series of usersettable
parameters, but the built-in defaults are reasonably
robust to many different protein/ligand pairs. All input molecules
must be protonated as expected at physiological pH
including non-polar hydrogens. The protonation state may
strongly affect docking. Docking requires a ligand, a protomol,
and a protein. Surflex will fragment the molecule, search
the fragments, dock the fragments, and construct the molecule
in the active site of the protein. The final output is the
top ten scoring conformations (Jain, A.N., 2003).
Docking Protocols
In the two algorithms studied here, the receptor is treated
as a rigid body and a grid potential is used to evaluate the
scoring functions. This simplification allows one to perform
docking more efficiently, which is especially crucial in database
screening.
Arguslab requires a PDB format file for both ligand and
receptor. The binding site was defined from the coordinates
of the ligand in the original PDB file. Argusdock exhaustive
search docking engine was used, with grid resolution of 0.40Å. Docking precision was set to ‘high precision’ and ‘flexible
ligand docking’ mode was employed for each docking
run.
In Surflex, the preferred input file format is Sybyl mol2.
Molecular output is generally in Sybyl mol2 format as well.
MDL mol or sd file (for ligands) and PDB (for proteins) are
also acceptable, although PDB files may generate errors or
unexpected results, since the format is frequently variable.
There are two scores for each docked conformation: an
affinity (-logKd) and a crash score (also pKd units). The
crash score is the degree of inappropriate penetration into
the protein by the ligand as well as the degree of internal
self-clashing that the ligand is experiencing (which is also
reported). Crash scores that are close to 0.0 are favorable.
The reported score includes the crash score reported (and
the crash score includes the self-clash term). The final
docked conformations are reported in descending order of
score, and correspond to the output files final-*.mol2. It’s
possible to optimize the fine position of a molecule that has
already been docked by a command and the optimized conformation
is stored in “opt.mol2” (Jain, A.N. 2004).
Results and Discussion
The best ranking poses predicted by the two programs
Arguslab and Surflex are shown in the figure 1 and their
root mean square deviation (RMSD) values from the original
crystallographic pose determined. It can be observed
that Surflex surpasses Arguslab. For RMSD interval < 2Å,
the difference in docking accuracies between the two programs
is so important but decrease significantly in RMSD
interval < 3Å.
Figure 2 shows the evaluation of the docking algorithms
for their sampling accuracy. The percentage of poses with
RMSD within 2Å from the experimental structure was 63%
for Surflex and only 33% for ArgusLab. This confirms the
results reported by earlier studies, Surflex appears to be
highly efficient in terms of sampling (Kellenberger, E. et al.,
2004). However, under less rigorous conditions, the performance
of ArgusLab is vastly improved with 62% of the top
ten poses falling within 3Å of the crystallographic pose. This
suggests that ArgusLab still gives some biological results
and can be used in educational demonstrations.
|
Figure 1: Best pose with reference to crystallographic pose.
|
|
Figure 2: Top ten poses with reference to crystallographic pose.
|
|
Figure 3: Ligand rotatable bonds in relation to docking accuracy.
|
|
Figure 4: % of hydrogen bonding in terms of docking accuracy.
|
Another kind of analysis we have carried out is the effect
of a ligand parameter on docking accuracy (figure 3). It is a well-known fact that as the number of rotatable bonds of
the ligand increases, the docking accuracy falls since a much
larger conformational space has to be sampled. The complexes
in the present study were divided into three groups,
ligands with 1 to <10 rotatable bonds, ligands with 11 to <
15 rotatable bonds and those with > 15 rotatable bonds. The
results confirm earlier works. Indeed for all algorithms, the
docking accuracy decreases when the number of rotatable
bonds increases. Also in all cases, accuracy of Surflex is
approximately double that of ArgusLab. This decrease is
very marked when the number of rotatable bonds exceed
15. Though, an essential remark is that docking time in
ArgusLab is typically much shorter than that of Surflex.
To further evaluate these docking programs, another test
we have conducted is to study the chemical nature of their
protein-ligand interactions and then to check the success
rate of each scoring function (figure 4). The classification
is aided by using X-Score. For any given protein-ligand complex,
if the contribution of the H-bond term in X-Score is
50% larger than the hydrophobic term, it is classified as the“hydrophilic” type. If the contribution of the hydrophobic
term is 50% larger than the H-bond term, it is classified as
the “hydrophobic” type. Otherwise, the complex is considered
to have mixed hydrophilic and hydrophobic factors in
the protein-ligand interaction and thus is classified as the“mixed” type. We have used X-Score for this classification
process because it is the only one with open source codes,
so we can analyze the hydrophobic and the hydrophilic terms
conveniently. Hydrogen bond driven complexes are the best
results given by Surflex (72%) and also for hydrophobicburial
driven ones (63%). There is no perceptible change in
the docking accuracy of ArgusLab with degree of hydrogen
bonding.
Studies for determination of IC50 and MIC, in specialized
laboratory, are needed to confirm these in silico results.
Conclusion
Our results prove that the Surflex program does a rational
job in docking and should assist significantly the drug
discovery process. Its use for molecular docking appears to
be most valuable. This study shows that the commercial
package surpasses the freely available docking program in
all parameters tested. The study also reveals that, in less
scrupulous conditions, ArgusLab can be used for demonstration
of molecular docking method to beginners in this
area owing to its easiness to use graphical user interface.
Moreover, some future advances can be made in this program
at the expense of the docking time.
References
-
ArgusLab 4.0, Mark AT Planaria Software LLC,
Seattle, http://www.ArgusLab.com.
- Bursulaya BD, Totrov M, Abagyan R, Brooks CL 3rd
(2003) Comparative study of several algorithms for
flexible ligand docking. J Comput Aided Mol Des 17:
755-763. » CrossRef
» PubMed » Google Scholar
- Halperin I, Ma B, Wolfson H, Nussinov R (2002)
Principles of docking: an overview of search algorithms
and a guide to scoring functions. Proteins 47: 409-443. » CrossRef
» PubMed » Google Scholar
- Jain AN (2003) Surflex: Fully Automatic Flexible Molecular
Docking Using a Molecular Similarity-Based
Search Engine. J Med Chem 46: 499-511. » CrossRef
» PubMed » Google Scholar
- Jain AN (2004) Virtual screening in lead discovery and
optimization. Current Opinion in Drug Discovery & Development
7: 396-403. » CrossRef » PubMed » Google Scholar
- Kellenberger E, Rodrigo J, Muller P, Rognan D (2004)
Comparative Evaluation of Eight Docking Tools for
Docking and Virtual Screening Accuracy. PROTEINS
Structure Function and Bioinformatics 57: 225-242. » CrossRef » PubMed » Google Scholar
- Kitchen DB, Decornez H, Furr JR, Bajorath J (2004)
Docking and scoring in virtual screening for drug discovery:
methods and applications. Nature reviews Drug
discovery 3: 935-949.
» CrossRef » PubMed » Google Scholar
- Lengauer T, Rarey M (1996) Computational methods
for biomolecular docking. Curr Opin Struct Biol 6:
402-406. » CrossRef » PubMed » Google Scholar
- Liu Z, Huang C, Fan K, Wei P, Chen H, et al. (2005)
Virtual screening of novel noncovalent inhibitors for
SARS-CoV 3C-like proteinase. J Chem Inf Model 45:
10-17. » CrossRef
» PubMed » Google Scholar
- Perola E, Walters WP, Charifson PS (2004) A detailed
comparison of current docking and scoring methods on
systems of pharmaceutical relevance. Proteins 56: 235-
249.
» CrossRef » PubMed » Google Scholar
- Sechi M, Sannia L, Carta F, Palomba M, Dallocchio R,
et al. (2005) Design of novel bioisosteres of beta-diketo
acid inhibitors of HIV-1 integrase. Antivir Chem
Chemother 16: 41-46. » CrossRef » PubMed » Google Scholar
- Taylor RD, Jewsbury PJ, Essex JW (2002) A review of protein-small molecule docking methods. J Comput Aided
Mol Des 16: 151-166. » CrossRef » PubMed » Google Scholar
- Thompson MA (2004) Molecular docking using ArgusLab, an efficient shape-based search algorithm and
the AScore scoring function. ACS meeting, Philadelphia
172 CINF 42 PA.
» Google Scholar
|
