Multiobjective Optimization
All the methods considered up to this point share a common element. They optimize a fitness function which comprises a set of weighted components. The effect is to turn the multiobjective problem into a single-objective optimization. There are a number of issues with this approach. The weights need to be defined empirically and may need to be varied from problem to problem. Whilst normalization of the properties may help this may in itself be nontrivial for certain properties. The very fact...
Identifying the Druggable Genome
The knowledge of which proteins current medicinal chemistry has developed drugs and leads against can be used to infer the subset of the proteins expressed by the human genome that have a high probability of being potentially druggable, i.e., capable of binding druglike small molecules with high affinity. The first systematic estimate of the number of druggable proteins - the 'druggable genome' - following the publication of the draft human genome,23'24 was based on a search for membership of...
References Xtv
1. Rozman, K. K. Doull, J. Toxicology 2001, 160, 191-196. 2. Crum-Brown, A. Fraser, T. R. Trans. Roy. Soc. Edinburgh 1868, 25, 151-203. 3. Richet, M. C. C. R. Soc. Biol. Paris 1893, 45, 775-776. 4. Meyer, H. Arch. Exp. Pathol. Pharmakol. 1899, 42, 109-118. 5. Overton, C. E. Gustav Fischer Jena Switzerland, 1901 English translation by Lipnick, R. L. Chapman and Hall London, 1991. 6. Albert, A. Goldacre, R. J. Phillips, J. J. Chem. Soc. 1948, 2240-2249. 7. Bell, P H. Roblin, R. O., Jr. J. Am....
Figure 21 Examples of the benzodiazepine scaffolds suggested as privileged in
compounds based on the b-turn motif and identified the first two nonpeptidic heterocyclic micromolar agonists associated with the melanocortin-1 receptor. Another example is the rapid identification of selective agonists of the five somatostatin-receptor subtypes through combinatorial chemistry.256 Rigid arrangement of atoms and the resulting large, fixed surface area of the important 'side chains' in peptide terms in the privileged scaffolds maximize the potential free energy of interaction,...
Compounds processed
Figure 13 Enrichment graphs for CDK2 virtual screens using a number of different fingerprint descriptors. The data highlight what can happen with compound collections that contain a large number of closely related analogues. When only overall hit enrichment is considered top graph Daylight fingerprints perform best. When Adjustments are made to only include novel chemotypes in the enrichment score bottom graph , Daylight fingerprint move to the bottom of the rankings. Adapted from Good, A. C....
Info Cuk
consequently a number of lead docking techniques have been devised to explicitly handle protein flexibility.278,279 Given the CPU limits imposed by VS calculations, however, such exhaustive approaches are typically not applicable to SVS, which generally treat the protein as rigid. While the incorporation of protein flexibility without the concomitant introduction of noise into the SVS continues to prove a difficult problem,233 a number of techniques have been developed that attempt to address...
Acknowledgments 1
This article is partially based on papers by Jacobson and Sali,177 Fiser and Sali,22 and Madhusudhan et al.221 We also acknowledge the funds from Sandler Family Supporting Foundation, NIH R01 GM54762, P01 GM71790, P01 A135707, and U54 GM62529, as well as Sun, IBM, and Intel for hardware gifts. 1. Congreve, M. Murray, C. W Blundell, T L. Drug Disc. Today 2005, 10, 895-907. 2. Hardy, L., Malikayil, A. Curr. Drug Disc. 2003, 15-20. 3. Lombardino, J. G. Lowe, J. A., III Nat. Rev. Drug Disc. 2004,...
Featurebased alignment
A number of techniques have been developed that exploit pharmacophoric feature constraints in order to direct ligand superposition calculations. While such calculations have been used for sometime in pharmacophore develop-ment,118,121,122 variants have also been proposed for explicit virtual screening calculations. The FLEXS system uses an incremental construction procedure detailed in Section 4.19.2.1.3 to build up a flexible test molecules onto a rigid reference structure. The program applies...
Table 4 Binding energy contributions for various functional groups as
Energy kcal 1 per functional group group to the observed binding affinity, together with a fixed entropy term 14kcalmol_related to the freezing of translational and rotational degrees of freedom of the molecule and a variable entropy term 0.3 kcal per bond summed over all rotatable bonds in the ligand. The values found are shown in Table 4. Although there is a large statistical variation in these values, Andrews proposed that summing the appropriate contributions for any novel ligand would give...
Info Bxs
pharmacophores present in many less active compounds simulated annealing
Layered Virtual Screening Application to Carbonic Anhydrase II and Checkpoint
In some cases structure-based VS can be part of a cascade of different computational techniques in the quest for binders of a given target. We give here two such examples. The first example of the combination of different VS techniques is given by a study that led to the successful identification of subnanomolar inhibitors of carbonic anhydrase II CAII .9'136 A layered strategy including pharmacophore and ligand-based modeling was applied, where the last step consisted of docking and scoring of...
Optimization Techniques
As indicated earlier, the computational intractability of exact methods for selecting the maximally diverse subset of compounds lead to the development of the DBCS and sphere exclusion approximate methods. An alternative approach is to use an optimization algorithm such as a genetic algorithm or simulated annealing. These algorithms can provide effective ways of sampling large search spaces and hence they are well suited to compound selection, provided that they are used with efficient methods...
Inhibiting pCateninTcf ProteinProtein Interactions8
b-catenin is an intracellular mediator of the Wnt signaling pathway.118'119 When Wnt signaling is activated, b-catenin together with Tcf proteins functions as a transcriptional activator of a large number of genes. The activation of some of these genes is essential for creating and maintaining the malignant phenotype of colorectal cancer cells.120,121 Consequently, the b-catenin-Tcf complex has emerged as an attractive anticancer drug target. The interaction between b-catenin and Tcf family...
Overview of De Novo Design
De novo ligand design has matured in the last 15-20 years and has been subject to several outstanding review articles.2-7 Figure 1 illustrates the conceptual steps in de novo design. Several problems make de novo ligand design difficult. First, estimation of binding energy with the target receptor is one problem that is common to methods of virtual screening as well. The rigorous physical chemical methods dealing with this question are often time-consuming and limited by the accuracy of the...
LigandBased Approaches Core Molecular Modeling
A R Leach, GlaxoSmithKline Research and Development, Stevenage, UK 2007 Elsevier Ltd. All Rights Reserved. 4.05.2 Molecular Mechanics and Empirical Force Field Methods 88 4.05.2.1 Bond Stretching and Compressing 89 4.05.2.3 Torsion Potentials 90 4.05.2.4 Out-of-Plane Bending Terms and Cross Terms 91 4.05.2.5 Nonbonded Interactions 92 4.05.2.6 Simple Models for Incorporating Electrostatic Interactions 92 4.05.2.8 The Dielectric Constant and Implicit Solvation Models 94 4.05.2.9 The Generalized...
Info Xes
Figure 18 The location of designed compounds red squares that were screening hits against a background of GPCR-PA ligands open circles . The view is the same as in Figure 17. a MCH-R hits b GnRH-R hits c MC4-R hits. Reprinted in part with permission from Lavrador, K. Murphy, B. Saunders, J. Struthers, S. Wang, X. Williams, J. J. Med. Chem. 2004, 47, 6864-6874. Copyright 2004 American Chemical Society. We have already highlighted the importance of the choice of chemistry and hence scaffold or...
O Hkl
H2N CCR3 IC50 2.3 nM CCR1 IC50 1 900 nM Figure 8 Implementation of descriptor-based library design in the development of a selective CCR3 antagonist. then ranked by their ability to differentiate biologically active compounds from inactive compounds. New scaffolds were then checked against the pharmacophores in the CRF1 design space. Informative design was applied to build the libraries. The computationally chosen cores, along with 12 amines that resulted from the informative design, were then...
H Ukt
Figure 5 Structure of the nicotinic analgesic ABT-594. Predicting the bioactive conformation of a given molecule, especially in the absence of structural information about the receptor, is an extremely difficult problem for molecular modeling in general due to the number of factors that contribute to ligand protein binding. In many cases, indirect methods like three-dimensional QSAR 3D-QSAR , which are always based on fundamental assumptions like for example that all molecules in a given data...
Subset Screening Analysis Iterative FollowUp and Triage
At some point, a subset will have been selected and distributed for use. What will be the recommended use, how will the screen be triaged, and how will it be followed up One way to approach these questions would be to define various general categories of screens and how they best fit into the subset screening scenarios. Once these have been defined, this information can serve as guidelines to prospective users of the subset to help manage expectations. The first scenario would be for projects...
Topological Descriptors 1
Two widely applied examples of 2D molecular descriptors are molecular connectivity indices MCI and atom pair AP descriptors, initially developed by Carhart etal.29 Most 2D QSAR methods have been extensively studied by Randic,32 and Kier and Hall 33-38 based on graph theoretic indices. Although the physicochemical meaning of these structural indices is unclear, they certainly represent different aspects of molecular structures. These topological indices have been successfully combined with MLR...
What Is Rajmol In Bio-informatics
1. Dean, P M. Computer-Aided Design of Small Molecules for Chemical Genomics. In Methods in Molecular Biology, Vol. 310, Chemical Genomics Reviews and Protocols Zanders, E. D., Ed. Humana Press Totowa, NJ, 2005, pp 25-39. 2. Lewis, R. A. Leach, A. R. J. Comput.-AidedMol. Des. 1994, 8, 467-475. 3. Murcko, M. A. Caron, P. R. Charifson, P S. Annu. Rep. Med. Chem. 1999, 34, 297-306. 4. Schneider, G. Clement-Chomienne, O. Hilfiger, L. Schneider, P. Kirsch, S. B hm, H.-J. Neidhart, W Angew. Chem....
Experimental Methods of Determining Partition Coefficients
The traditional shake-flask method, albeit slow and laborious, has been used to reliably measure many thousands of precise octanol water partition coefficients.46'47 An excellent analysis of this method is described by Taylor.48 However, the time and resources required for this methodology limits its effectiveness, particularly with the advent of combinatorial synthesis and generation of diverse libraries. The widespread application of partition coefficients in ADME, toxicology, environmental...
Hydrophobicity to the Rescue
One of the earliest attempts to describe biological interactions in a quantitative sense was made by Hansen in 1962.41 He examined the toxicity of a series of substituted benzoic acids in terms of their a values and obtained a reasonable but fortuitous correlation as was later realized.42 Against this backdrop, in the early 1960s we 'plodded' our way through our studies based on the interactions of x-phenoxyacetic acids with oat shoots. We were fortunate to have Robert Muir, a biologist with...
Design Considerations for Libraries of all Sizes and Types
When designing a combinatorial or parallel library, it is essential to consider the intended use of the library. The library size and makeup of a general screening library are different from that of a target family library, which is, in turn, different from a library intended for a single target or for optimization of a lead. Clearly, the largest library type is the general screening or corporate deck enhancement library. These are most likely to be based on some definition of molecular...
Summary and Conclusions
In this chapter we have surveyed a broad range of techniques that are widely used in computational drug discovery. Some of these methods trace their origins to the earliest days of computational molecular science, and continue to present challenges to the computational chemistry community. As computer power has increased so too has the desire to study larger and more complex systems, or to apply the methods to much larger collections of compounds in virtual screening experiments. New force...
StructureBased Methods for Combinatorial Libraries
In its simplest form, structure-based library design is merely a variation of well-established virtual screening techniques as described in Chapter 4.19 and reviewed by Bohm and Stahl126 and Beavers and Chen.127 However, enumeration and docking of an entire library are inefficient and, because of the significant time required, impractical for large libraries. Several methods have been reported recently that address this problem by taking advantage of the modular nature of combinatorial...
Techniques to Describe the Molecule and its Pharmacophore Features to the
Table 2 lists programs that use special enhancements to conformational searching that enable one to propose one or more potential bioactive conformations of molecules that interact with the same macromolecular binding site. These approaches are especially appropriate to use when one has a firm idea or hypothesis of the points that should match and when the molecules are quite flexible. To view a molecular structure on the computer screen one represents the 3D structure as a set of Cartesian,...
Py y2
where y,, y,, and y are the actual activities, the estimated activities by LOO cross-validation procedure, and the average activities, respectively. The summations in eqn 1 are performed over all compounds used to build a model i.e., the training set . Frequently, q2 is used as the criterion of both robustness and predictive ability of the model. Many authors consider high q2 for instance, q2 gt 0.5 as an indicator or even as the ultimate proof of the high predictive power of a QSAR model. They...
log Ki b log P b[26
He suggested that if b 0.5 0.2, a ligand would bind on the surface of a receptor and if b 1 0.2, it would be transferred into the interior of a hydrophobic pocket.93 Extensive QSAR analyses from the Hansch laboratory were in agreement with these observations.94 Ten years later, the advent of molecular graphics and the availability of x-ray crystal structures of adequate resolution allowed us to clearly link the coefficients of the hydrophobic parameter with the extent of desolvation at the...
Hansch Approach
In 1964, after analyzing a large number of data sets, Hansch and Dunn formulated a more generalized approach to SARs. The critical breakthrough was the usage of a linear combination of different physicochemical parameters to delineate biological activity.78 log 1 C alog P pa constant 21 C is the molar concentration that evokes a certain biological response. The utility of this model which was developed by using multiple linear regression MLR analysis has been demonstrated in all types of...
Limitations of the Assumptions for Pharmacophore Mapping
Because of unavailability of experimental 3D structures of the target biomolecule, one may have no choice but to use pharmacophore-mapping methods to gain insight into the 3D features required for bioactivity. However, protein crystallography studies illustrate the artificiality of the underlying assumptions of pharmacophore mapping. There is abundant evidence that proteins change their structure to accommodate ligands and that this change in structure is not the same for all ligands, even...
Screening by crystallography
The strategy of directly obtaining x-ray crystallographic data on small fragments bound to proteins is well documented.28,32 A clear advantage of this approach is that false positives are reduced because if a compound is seen by crystallography then an immediate assessment can be made of how to enhance binding by using modeling techniques. On the downside, these techniques can be very time and resource intensive in that a large milligram amount of a protein construct is needed that is...
Acknowledgments
Financial support from the Klaus Tschira Foundation is gratefully acknowledged. 1. Fischer, E. Ber. Deutsch Chem. Ges. 1894, 27, 2985-2993. 2. Burgen, A. S. V. Roberts, G. C. K. Feeny, J. Nature 1975, 253, 753-755. 3. Koshland, D. E., Jr. Proc. Natl. Acad. Sci. USA 1958, 44, 98-123. 4. J Tsai, C. Ma, B. Nussinov, R. Proc. Natl. Acad. Sci. USA 1999, 96, 9970-9972. 5. Teague, S. J. Nat. Rev. Drug Disc. 2003, 2, 527-541. 6. Schames, J. R. Henchman, R. H. Siegel, J. S. Sotriffer, C. A. Ni, H....
General Principles
Ligand-receptor interactions in vivo are determined by a number of diverse factors. The first issue to consider is whether the ligand and receptor can approach close enough to recognize and bind to each other. Does a ligand have to cross a membrane to reach the same cellular compartment as a receptor How will an orally administered drug reach the bloodstream and then its target receptor To answer these questions, the physicochemical properties of the ligand need to be considered as well as its...
Pose Prediction
In many cases it is necessary to evaluate docking programs with respect to their ability to reproduce experimentally known poses in a reliable manner. The traditional way to do this is to calculate the RMS deviation RMSD between a pose generated by a docking program and the experimentally observed binding mode. Despite the practical appeal of using RMSDs from a crystal structure to assess pose prediction accuracy, they do not do justice to the complex interactions ligands make with proteins....
Info Ghf
The steric parameters developed up to this point focused mainly on describing intramolecular steric hindrance in organic reactions. Increased emphasis on intermolecular interactions led to the development of the STERIMOL parameters L, Bj to B4 by Verloop etal., a few years later.32'33 They based their parameters on standard bond angles and bond distances L represented the length of the substituent while Bi-B4 represented various perpendicular widths in four different directions. While this...
Evolution of other Shape Descriptors
Verloop's STERIMOL descriptors focused specifically on substituent shape. Simon etal. chose to examine the shape of the whole molecule by employing 'superimposition' techniques to determine the minimal steric difference MSD and the minimum topological difference MTD . They defined 'hypermolecules' which were then used with hydrophobic and electronic parameters in regression analysis.110,111 Treatment of the hypermolecules as rigid entities, limited the utility of this approach. Hopfinger's...
Info Cyk
aThe designed set used a five-dimensional chemistry space to design libraries based on the compound adjacency to known GPCR-PA ligands. The random and basic sets were selected at random from a screening collection, with the latter set constrained to compounds containing a basic group. aThe designed set used a five-dimensional chemistry space to design libraries based on the compound adjacency to known GPCR-PA ligands. The random and basic sets were selected at random from a screening...
References Nbe
1. Burbaum, J. J. Ohlmeyer, M. H. Reader, J. C. Henderson, I. Dillard, L. W Li, G. Randle, T L. Sigal, N. H. Chelsky, D. Baldwin, J. J. Proc. Natl. Acad. Sci. USA 1995, 92, 6027-6031. 2. Blondelle, S. E. Crooks, E. Ostresh, J. M. Houghten, R. A. Antimicrob. Agents Chemother. 1999, 43, 106-114. 3. Boger, D. L. Jiang, W Goldberg, J. J. Org. Chem. 1999, 64, 7094-7100. 4. Ferry, G. Boutin, J. A. Atassi, G. Fauchere, J. L. Tucker, G. C. Mol. Div 1996, 2, 135-146. 5. Lutzke, R. A. P Eppens, N. A....
Docking Programs Quick Explore QXP
QXP Quick Explore17 is part of the Flo program. It contains two conceptually different docking algorithms MCDock and ZipDock. The MCDock algorithm is evolutionary in nature and similar to the conformational space annealing method proposed by Lee and Scheraga.31 For each ligand it applies a user-defined number of repeated cycles of Monte Carlo followed by energy minimization to generate and refine an ensemble of low-energy ligand poses. By adding dissimilar, low-energy poses to the ensemble and...
Topological indices
Topological indices have been widely used in various studies. A topological index is a single value that can be calculated from the 2D graph representation of a molecule.37'38 Many hundreds of different indices have been described in the literature. The simplest topological indices are based on characterizing structures according to their size, degree of branching, and overall shape. More complex indices include consideration of the properties of atoms as well as their connectivities. The...
Qsar
4.06 Pharmacophore Modeling 1 - Methods 4.21 Pharmacophore Modeling 2 - Applications 4.02 Introduction to Computer-Assisted Drug Design - Overview and Perspective for the Future 4.03 Quantitative Structure-Activity Relationship - A Historical Perspective and the Future 4.07 Predictive Quantitative Structure-Activity Relationship Modeling 4.22 Topological Quantitative Structure-Activity Relationship Applications Structure Information Representation in Drug Discovery 4.23 Three-Dimensional...
References Ezt
1. Willett, P Similarity and Clustering in Chemical Information Systems Research Studies Press Letchworth, 1987. 2. Johnson, M. A. Maggiora, G. M. Concepts and Applications of Molecular Similarity John Wiley New York, 1990. 3. Downs, G. M. Willett, P Rev. Comput. Chem. 1995, 7, 1-66. 4. Sheridan, R. P. Kearsley, S. K. Drug Disc. Today 2002, 7, 903-911. 5. Nikolova, N. Jaworska, J. Quant. Struct.-Act. Relat. Comb. Sci. 2003, 22, 1006-1026. 6. Dean, P. M. Molecular Similarity in Drug Des. Chapman...
References Wtm
1. B hm, H.-J. Schneider, G., Eds. Virtual Screening for Bioactive Molecules Wiiey-VCH Weinheim, 2000 Vol. 10. 2. Schneider, G. B hm, H.-J. Drug Disc. Today 2002, 7, 64-70. 3. Waszkowycz, B. Curr. Opin. Drug Disc. Dev. 2002, 5, 407-413. 4. Toledo-Sherman, L. M. Chen, D. Curr. Opin. Drug Disc. Dev. 2002, 5, 414-421. 5. Verkhivker, G. M. Bouzida, D. Gehlhaar, D. K. Rejto, P. A. Arthurs, S. Colson, A. B. Freer, S. T Larson, V Luty, B. A. Marrone, T. et al. J. Comput-Aided Mol. Design. 2000, 14,...
DissimilarityBased Compound Selection
DBCS methods involve selecting a subset of compounds directly based on their pairwise dissimilarities. They are iterative procedures whereby one compound is selected in each iteration. The basic algorithm for DBCS is outlined below 1. Select a compound and place it in the subset. 2. Calculate the dissimilarity between each compound remaining in the dataset and the compounds in the subset. 3. Choose the next compound as that which is most dissimilar to the compounds in the subset. 4. If there...
Table 2 Descriptions of some distance metrics and similarity coefficients
Formula for continuous variables Formula for dichotomous variables Set-theoretic definition Range Formula for continuous variables Formula for dichotomous variables Normalized complement for dichotomous data, called simple matching coefficient to 0 continuous , n to 0 dichotomous Obeys all four metric properties Equivalent to the squared Euclidean distance for dichotomous variables Can be normalized to the range 1-0 if the values of all attributes are normalized to this range and the result...
References Lpq
1. Gund, P. Prog. Mol. Subcell. Biol. 1977, 5, 117-143. 2. Kier, L. B. Molecular Conformation. Molecular Orbital Theory in Drug Research Academic Press New York, 1971, pp 164-195. 3. Dror, O. Shulman, P A. Nussinov, R. Wolfson, H. J. Curr. Med. Chem. 2004, 11, 71-90. 3a. Van Drie, J. H. Curr. Pharm. Des. 2003, 9, 1649-1664. 4. Gtiner, O. F Pharmacophore Perception, Development, and Use in Drug Design International University Line La Jolla, CA, 1999, p. 537. 5. Cottrell, S. J. Gillet, V J....
References 1
1. Allen, E H. Motherwell, W D. Acta Crystallogr., Sect. B 2002, 58, 407-422. 2. Allen, E H. Davies, J. E. Galloy, J. J. Johnson, O. Kennard, O. Macrea, C. E Mitchell, E. M. Mitchell, G. E Smith, J. M. Watson, D. G. J. Chem. Inf Comput. Sci. 1991, 31, 187-204. 3. Abola, E. E. Bernstein, E C. Koetzle, T. E The Protein Data Bank. In The Role of Data in Scientific Progress Glaeser, P. S., Ed. Elsevier New York, 1985. 4. Eujita, T Iwasa, J. Hansch, C. J. Am. Chem. Soc. 1964, 86, 5175-5180. 5....
Info Mdj
side chains past the Cp atom. The details of atomic resolution are avoided because of the precision required to pack side chains efficiently without van der Waals overlap. As the quality of fold predictions increases, however, the ability to discriminate between alternative folds becomes more difficult and requires a high-resolution force field including multipole electrostatics and polarization. A recent report from the Baker group suggests high-resolution structures lt 1.5 A RMSD can be...
StructureBased Drug Design Examples
4.04.2.1 Hemoglobin The First Drug Design Target Perutz and colleagues, starting with the first low-resolution structures in the late 1950s, took 25 years to solve the highresolution structures of oxy and deoxy horse 1968, 1970 and human Hb 19 83, 19 84 .70-73 Perutz and co-workers subsequently determined the structures of a number of mutant Hb that caused different medical maladies.74-86 The solved structures readily explained at the molecular level the reason for the physical impairments...