Faulon J.L., Bender A. Handbook of Chemoinformatics Algorithms
Подождите немного. Документ загружается.
308 Handbook of Chemoinformatics Algorithms
10.3.3 SAMPLING STRATEGIES WITH EAs
Sampling strategies differ from grow and fragment-link in that they sample structures
without directing generation in a particular direction (outward or explicitly linking
interaction sites). EAs are the most common chosen for this purpose. There are many
types of EAs: genetic algorithms (GAs) that encode the molecular structure in a
“chromosome” of fixed length that is operated on and transformed into the molecular
structure for fitness evaluations; genetic programming, where the chromosomes are
trees to allow them to have variable length; and evolutionary strategies that operate
directly on the phenotype, which is the molecular structure. A basic evolutionary
strategy (μ, λ) is shown below [63], where λ is the size of the child population and
μ is the size of the parent population in each generation.
The algorithm starts with a population of λ chemical structures (the initial child
population) generated from putting a random selection of building blocks together
according to the building rules for the building blocks. Each structure in this population
is evaluated for “fitness.” The fitness is the scoring function that can combine primary
and secondary constraints. In receptor-based de novo methods, the primary score may
be the interaction score such as from a docking calculation [64], minus any boundary
violations. For ligand-based employing of a single template ligand, the primary score
may be a similarity score. Secondary scoring considerations, such as requirements
for Lipinski’s [57] rules, or other ADMET considerations, can be added to the fitness
function here, along with other molecule properties such as surface area or radius of
gyration. The most fit μ structures are selected to be parents for the next generation.
Mutation and crossover operators can be performed on the parents. Mutation in this
case is to take a parent structure and remove a building block and replace it according
to the joining rules. Crossover is to remove building blocks from each parent and
swap them again according to joining rules. Some algorithms have only mutation
[24] or only crossover [52]. A total of λ child structures are generated. This cycle is
repeated until it reaches a maximum generation of children or a termination condition
is reached, such as convergence. One feature found with this algorithm was that since
building blocks could vary greatly in size, the parent p could grow and shrink as well,
while still retaining the same number of building blocks.
ALGORITHM 10.3: PSEUDOCODE FOR BASIC EVOUTIONARY
STRATEGY (μ, λ)
## μ is the size of parent population
## λ is the size of the child population
Generate λ random structures Sc
DO
Evaluate fitness Fc of each structure Sc in
population
Choose μ most fit (Fc) structures as parents Sp
Mutate and Crossover of parents Sp to generate
population λ children Sc
UNTIL (> maximum generations or termination condition)
Computer-Aided Molecular Design 309
10.3.3.1 Programs in Current Use that Implement Sampling Strategies
EAs are common especially in ligand-based design programs, although several
receptor-based programs also employ this approach. See Table 10.1.c for some exam-
ples of lead compounds identified using the EA. One successful implementation of
this algorithm is in the TOPAS [24] ligand-based de novo program, which uses pair-
wise similarity to a molecular template as the fitness function. It sets λ = 100 and
μ = 1 (i.e., 1 parent) with no crossover operation, so all variations are through muta-
tion. It uses 25,000 fragments from the WDI using 11 retrosynthetic pathways. The
variance in each new child structure can be controlled by how similar a new building
block is to the original building block being mutated, and is controlled by a parameter
(“step-size”) that is a Gaussian distribution of random numbers, resulting in a child
population that is bell-shaped distribution of variations with the parent at the center.
It was found that 100 generations were sufficient to explore chemical space in this
program. TOPAS is at Hoffmann-La Roche and is not generally available.
Flux [37] was developed based on TOPAS. It finds optimal results with a 50:50
ratio between crossover and mutation, and typically sets maximum generations to 75
(50 generations were found to converge in most cases). The other main difference
from TOPAS is a modified similarity descriptor that is weighted. It is being used at
the Goethe University in Germany but is currently not generally available.
LEA3D uses fragments as building blocks generating from fragmenting “drug-
like” database of over 8,000 fragments into single rings, fused rings, and acyclic parts.
It allows both 1 and 2 point crossover and mutation. It is not generally available, but an
in-house versionis in use at the Centre De Biochimie Structurale, Montepelier,France.
De novo programs incorporating EAs that are commercially available include
AutoLudi and LeapFrog [48] and the Molecule Evoluator [56,65]. AutoLudi is an
extension of LUDI that uses an EA to modify an existing lead compound by adding on
small fragments. LeapFrog commercially available at Tripos (http://www.tripos.com)
evolves a population of molecules in an atom-based method. The Molecular Evolu-
ator [56] uses an unusual fitness function—the user working interactively with the
program. It is available at Cidrux Pharmaceuticals [66].
10.3.3.2 Advantages, Limitations, and Computational Complexity?
A general challenge for EAs is the molecule representation. SMILES strings such as
in LEA [11] have the problem that invalid molecules result during crossover and muta-
tion, and also more steps are required to build up a molecule. TreeSmiles, a variation
of SMILES with all hydrogens explicitly shown, helps avoid unreasonable structures
[56]. The LEA3D successor of LEA uses fragments instead of atoms as the build-
ing block, with numbers representing fragments for genetic operations [67]. Other
approaches operate directly on the 3D structures leading to additional translational
and rotational operators [36].
The theoretical chemical space available for a fragment-based approach is (nb)
ns
,
where nb is the number of building blocks in the fragment library, and ns is the
average number of building blocks is the final structure. For TOPAS that has 25,000
building blocks and approximately four building blocks in a final structure, the total
is (25,000)
4
≈ 10
18
. For an atom-based method, this would approach all of chemical
310 Handbook of Chemoinformatics Algorithms
space (≈10
60
). However, the number of structures actually evaluated in an EA run
is much smaller as it is given by the function λ ng, where λ represents the popula-
tion of each generation and ng the number of generations. For example, TOPAS has
population size 100 ×100 generations ≈10
4
. Similarly, LEA3D has a a population
size of 40 ×100 generations, that is, 4000. In practice, this seems to be sufficient to
generate enough reasonable solutions to find interesting leads.
Compared to the grow and fragment-link, EA algorithms have the advantage that,
since they do not target interaction site points, the output structures are not strained
(i.e., have low intramolecular energies). The corresponding disadvantage is that they
may not bind to known important interaction sites.
10.4 SUMMARY
In examining all grow, fragment-link, and sampling-based algorithms, one aspect in
common is the use of a random operator of some sort during the structure generation
process. This is important for two reasons: first because the scoring functions are not
perfect; the best-scoring atom or fragment may not represent the best binder. Second,
and more importantly, because the path to construct a de novo structure is not a linear
function of the scoring function (i.e., higher scoring final structures are not a linear
result of the highest scoring precursors; a structure often needs to go through a lower
energy construction pathway to get to the final structure).
Ligand-based programs that use similarity to a molecular template or QSAR for
scoring require a sampling approach, and the EA is the most commonly chosen
one for these programs for its simplicity to program up and its effectiveness in these
cases. With receptor-based approaches, the grow and fragment-link algorithms, which
include pharmacophore data in the form of site points, are historically favored. Pure
sampling approaches are most commonly seen as lead-optimization once a core has
been designed using a grow or fragment-link approach.
The major hurdles for de novo design to overcome to be an effective tool in drug
discovery are the same today as when the field began: how to accurately predict
receptor-based affinity and predict synthetic accessibility. Without these, it could be a
costly effort to synthesize complex molecules, which may not even bind to the target
receptor. Newer heuristics for synthetic accessibility, using reaction-based fragment
libraries, and heuristics based on molecular complexity have improved the quality of
structures resulting from de novo design. Several de novo design strategies have now
been shown to be successful in prospective studies.
REFERENCES
1. Bohacek, R. S., McMartin, C., and Guida, W. C., The art and practice of structure-based
drug design: A molecular modeling perspective. Med. Res. Rev. 1996, 16(1), 3–50.
2. Anonymous, The numbers game. Nat. Rev. Drug Discov. 2002, 1(12), 929.
3. Bolton, E. E., Wang, Y., Thiessen, P. A., and Bryant, S. H., PubChem: Integrated platform
of small molecules and biological activities, Annual Reports in Computational Chemistry,
2008, p. 4.
Computer-Aided Molecular Design 311
4. Danziger, D. J. and Dean, P. M., Automated site-directed drug design: A general algorithm
for knowledge acquisition about hydrogen-bonding regions at protein surfaces. Proc. R.
Soc. Lond. B Biol. Sci. 1989, 236(1283), 101–113.
5. Lewis, R. A. and Dean, P. M., Automated site-directed drug design: The formation of
molecular templates in primary structure generation. Proc. R. Soc. Lond. B. Biol. Sci.
1989, 236(1283), 141–162.
6. Lewis, R. A. and Dean, P. M., Automated site-directed drug design: The concept of
spacer skeletons for primary structure generation. Proc. R. Soc. Lond. B Biol. Sci. 1989,
236(1283), 125–140.
7. Lewis, R. A., Roe, D. C., Huang, C., Ferrin, T. E., Langridge, R., and Kuntz, I. D., Auto-
mated site-directed drug design using molecular lattices. J. Mol. Graph. 1992, 10(2),
66–78, 106.
8. Lewis, R. A., Automated site-directed drug design: Approaches to the formation of 3D
molecular graphs. J. Comput. Aided Mol. Des. 1990, 4(2), 205–210.
9. Gillet, V. J., Newell, W., Mata, P., Myatt, G., Sike, S., Zsoldos, Z., and Johnson, A. P.,
SPROUT: Recent developments in the de novo design of molecules. J. Chem. Inf. Comput.
Sci. 1994, 34(1), 207–217.
10. Clark, D. E., Frenkel, D., Levy, S. A., Li, J., Murray, C. W., Robson, B., Waszkowycz, B.,
and Westhead, D. R., PRO-LIGAND: An approach to de novo molecular design. 1. Appli-
cation to the design of organic molecules. J. Comput. Aided Mol. Des. 1995, 9(1), 13–32.
11. Douguet, D., Thoreau, E., and Grassy, G., A genetic algorithm for the automated genera-
tion of small organic molecules: Drug design using an evolutionary algorithm. J. Comput.
Aided Mol. Des. 2000, 14(5), 449–466.
12. Schneider, G., Clement-Chomienne, O., Hilfiger, L., Schneider, P., Kirsch, S., Bohm, H.
J., and Neidhart, W., Virtual screening for bioactive molecules by evolutionary de novo
design. Angew Chem. Int. Ed. Engl. 2000, 39(22), 4130–4133.
13. Goodford, P. J., A computational procedure for determining energetically favorable bind-
ing sites on biologically important macromolecules. J. Med. Chem. 1985, 28(7), 849–857.
14. Bohm, H. J., The computer program LUDI: A new method for the de novo design of
enzyme inhibitors. J. Comput. Aided Mol. Des. 1992, 6(1), 61–78.
15. Rotstein, S. H. and Murcko, M. A., GroupBuild: A fragment-based method for de novo
drug design. J. Med. Chem. 1993, 36(12), 1700–1710.
16. Bohm, H. J., LUDI: Rule-based automatic design of new substituents for enzyme inhibitor
leads. J. Comput. Aided Mol. Des. 1992, 6(6), 593–606.
17. Gillet, V., Myatt, G., Zsoldos, Z., and Johnson, A., SPROUT, HIPPO and CAESA: Tools
for de novo structure generation and estimation of synthetic accessibility. Perspect. Drug
Discov. Des. 1995, 3(1), 34–50.
18. Miranker, A. and Karplus, M., Functionality maps of binding sites: A multiple copy
simultaneous search method. Proteins 1991, 11(1), 29–34.
19. Roe, D. C. and Kuntz, I. D., BUILDER v.2: Improving the chemistry of a de novo design
strategy. J. Comput. Aided Mol. Des. 1995, 9(3), 269–282.
20. Gillet, V., Johnson, A. P., Mata, P., Sike, S., and Williams, P., SPROUT: A program for
structure generation. J. Comput. Aided Mol. Des. 1993, 7(2), 127–153.
21. Todorov, N. P. and Dean, P. M., A branch-and-bound method for optimal atom-type
assignment in de novo ligand design. J. Comput. Aided Mol. Des. 1998, 12(4), 335–410.
22. Nishibata, Y. and Itai, A., Automatic creation of drug candidate structures based on
receptor structure. Starting point for artificial lead generation. Tetrahedron 1991, 47(43),
8985–8990.
23. Lewell, X. Q., Judd, D. B., Watson, S. P., and Hann, M. M., RECAP—retrosynthetic
combinatorial
analysis
procedure: A powerful new technique for identifying privileged
312 Handbook of Chemoinformatics Algorithms
molecular fragments with useful applications in combinatorial chemistry. J. Chem. Inf.
Comput. Sci. 1998, 38(3), 511–522.
24. Schneider, G., Lee, M. L., Stahl, M., and Schneider, P., De novo design of molecular
architectures by evolutionary assembly of drug-derived building blocks. J. Comput. Aided
Mol. Des. 2000, 14(5), 487–494.
25. Eisen, M. B., Wiley, D. C., Karplus, M., and Hubbard, R. E., HOOK: A program for
finding novel molecular architectures that satisfy the chemical and steric requirements of
a macromolecule binding site. Proteins 1994, 19(3), 199–221.
26. Degen, J. and Rarey, M., FlexNovo: Structure-based searching in large fragment spaces.
Chem. Med. Chem. 2006, 1(8), 854–868.
27. Wang, R., Gao, Y., and Lai, L., LigBuilder: A multi-purpose program for structure-based
drug design. J. Mol. Model. 2000, 6(7), 498–516.
28. Moon, J. B. and Howe, W. J., Computer design of bioactive molecules: A method for
receptor-based de novo ligand design. Proteins 1991, 11(4), 314–328.
29. Rotstein, S. H. and Murcko, M. A., GenStar:A method for de novo drug design. J. Comput.
Aided Mol. Des. 1993, 7(1), 23–43.
30. Bohacek, R. S. and McMartin, C., Multiple highly diverse structures complementary
to enzyme binding sites: Results of extensive application of a de novo design method
incorporating combinatorial growth. J. Am. Chem. Soc. 1994, 116(13), 5560–5571.
31. Murray, C. W., Clark, D. E., Auton, T. R., Firth, M. A., Li, J., Sykes, R. A., Waszkowycz,
B., Westhead, D. R., and Young, S. C., PRO_SELECT: Combining structure-based drug
design and combinatorial chemistry for rapid lead discovery. 1. Technology. J. Comput.
Aided Mol. Des. 1997, 11(2), 193–207.
32. David, A. and Pearlman, M. A. M., CONCEPTS: New dynamic algorithm for de novo
drug suggestion. J. Comput. Chem. 1993, 14(10), 1184–1193.
33. Gehlhaar, D. K., Moerder, K. E., Zichi, D., Sherman, C. J., Ogden, R. C., and Freer, S. T.,
De novo design of enzyme inhibitors by Monte Carlo ligand generation. J. Med. Chem.
1995, 38(3), 466–472.
34. Todorov, N. P. and Dean, P. M., Evaluation of a method for controlling molecular scaffold
diversity in de novo ligand design. J. Comput. Aided Mol. Des. 1997, 11(2), 175–192.
35. Hartenfeller, M., Proschak, E., Schuller, A., and Schneider, G., Concept of combinatorial
de novo design of drug-like molecules by particle swarm optimization. Chem. Biol. Drug
Des. 2008, 72(1), 16–26.
36. Glen, R. C. and Payne, A. W., A genetic algorithm for the automated generation of
molecules within constraints. J. Comput. Aided Mol. Des. 1995, 9(2), 181–202.
37. Fechner, U. and Schneider, G., Flux (2): Comparison of molecular mutation and crossover
operators for ligand-based de novo design. J. Chem. Inf. Model. 2007, 47(2), 656–667.
38. Vinkers, H. M., de Jonge, M. R., Daeyaert, F. F., Heeres, J., Koymans, L. M., van Lenthe, J.
H., Lewi, P. J., Timmerman, H., Van Aken, K., and Janssen, P. A., SYNOPSIS: Synthesize
and optimize system in silico. J. Med. Chem. 2003, 46(13), 2765–2773.
39. Pierce, A. C., Rao, G., Bemis, G. W., BREED: Generating novel inhibitors through
hybridization of known ligands. Application to CDK2, p38, and HIV protease. J. Med.
Chem. 2004, 47(11), 2768–2775.
40. Nachbar, R. B., Molecular evolution: Automated manipulation of hierarchical chemical
topology and its application to average molecular structures. Genet. Prog. Evolv. Mach.
2000, 1(1–2), 57–94.
41. Andrew, R. and Leach, R. A. L., A ring-bracing approach to computer-assisted ligand
design. J. Comput. Chem. 1994, 15(2), 233–240.
Computer-Aided Molecular Design 313
42. Bohm, H. J., Prediction of binding constants of protein ligands: A fast method for the
prioritization of hits obtained from de novo design or 3D database search programs.
J. Comput. Aided Mol. Des. 1998, 12(4), 309–323.
43. Bohm, H. J., The development of a simple empirical scoring function to estimate the
binding constant for a protein–ligand complex of known three-dimensional structure.
J. Comput. Aided Mol. Des. 1994, 8(3), 243–256.
44. Ishchenko, A. V. and Shakhnovich, E. I., SMall molecule growth 2001 (SMoG2001):
An improved knowledge-based scoring function for protein–ligand interactions. J. Med.
Chem. 2002, 45(13), 2770–2780.
45. DeWitte, R. S. and Shakhnovich, E. I., SMoG: De novo design method based on simple,
fast, and accurate free energy estimates. 1. Methodology and supporting evidence. J. Am.
Chem. Soc. 1996, 118(47), 11733–11744.
46. Todorov, N. P., Buenemann, C. L., and Alberts, I. L., De novo ligand design to an ensemble
of protein structures. Proteins 2006, 64(1), 43–510.
47. Alberts, I. L., Todorov, N. P., and Dean, P. M., Receptor flexibility in de novo ligand design
and docking. J. Med. Chem. 2005, 48(21), 6585–6596.
48. Leapfrog, SYBYL 7.1, TRIPOS: St. Louis, MO.
49. Waszkowycz, B., Clark, D. E., Frenkel, D., Li, J., Murray, C. W., Robson, B., and West-
head, D. R., PRO_LIGAND: An approach to de novo molecular design. 2. Design of
novel molecules from molecular field analysis (MFA) models and pharmacophores. J.
Med. Chem. 1994, 37(23), 3994–4002.
50. Pellegrini,E. and Field, M. J., Development and testing of a de novo drug-design algorithm.
J. Comput. Aided Mol. Des. 2003, 17(10), 621–641.
51. Brown, N., McKay, B., Gilardoni, F., and Gasteiger, J., A graph-based genetic algorithm
and its application to the multiobjective evolution of median molecules. J. Chem. Inf.
Comput. Sci. 2004, 44(3), 1079–1087.
52. Globus, A., Lawton, J., and Wipke, T., Automatic molecular design using evolutionary
techniques. Nanotechnology 1999, 10(3), 290–299.
53. Honma, T., Hayashi, K., Aoyama, T., Hashimoto, N., Machida, T., Fukasawa, K.,
Iwama, T., et al., Structure-based generation of a new class of potent Cdk4 inhibitors:
New de novo design strategy and library design. J. Med. Chem. 2001, 44(26),
4615–4627.
54. Boda, K. and Johnson, A. P., Molecular complexity analysis of de novo designed ligands.
J. Med. Chem. 2006, 49(20), 5869–5879.
55. Gasteiger, J., De novo design and synthetic accessibility. J. Comput. Aided Mol. Des. 2007,
21(6), 307–310.
56. Lameijer, E. W., Kok, J. N., Back, T., and Ijzerman, A. P., The molecule evoluator. An
interactive evolutionary algorithm for the design of drug-like molecules. J. Chem. Inf.
Model. 2006, 46(2), 545–552.
57. Lipinski, C. A., Lombardo, F., Dominy, B. W., and Feeney, P. J., Experimental and
computational approaches to estimate solubility and permeability in drug discovery and
development settings. Adv. Drug Deliv. Rev. 2001, 46(1–3), 3–26.
58. Aronov, A. M., Predictive in silico modeling for hERG channel blockers. Drug Discov.
Today 2005, 10(2), 149–155.
59. Kick, E. K., Roe, D. C., Skillman, A. G., Liu, G., Ewing, T. J., Sun, Y., Kuntz, I. D., and
Ellman, J. A., Structure-based design and combinatorial chemistry yield low nanomolar
inhibitors of cathepsin D. Chem. Biol. 1997, 4(4), 297–307.
60. Ripka, A. S., Satyshur, K. A., Bohacek, R. S., and Rich, D. H., Aspartic protease inhibitors
designed from computer-generated templates bind as predicted. Org. Lett. 2001, 3(15),
2309–2312.
314 Handbook of Chemoinformatics Algorithms
61. Caflisch, A., Miranker, A., and Karplus, M., Multiple copy simultaneous search and con-
struction of ligands in binding sites:Application to inhibitors of HIV-1 aspartic proteinase.
J. Med. Chem. 1993, 36(15), 2142–2167.
62. Dong, X., Zhang, Z., Wen, R., Shen, J., Shen, X., and Jiang, H., Structure-based de novo
design, synthesis, and biological evaluation of the indole-based PPARgamma ligands (I).
Bioorg. Med. Chem. Lett. 2006, 16(22), 5913–5916.
63. Schwefel, H. P., Deep insight from simple models of evolution. Biosystems 2002, 64(1–3),
189–198.
64. Pegg, S. C., Haresco, J. J., and Kuntz, I. D., A genetic algorithm for structure-based de
novo design. J. Comput. Aided Mol. Des. 2001, 15(10), 911–933.
65. Lameijer, E.W., Tromp, R.A., Spanjersberg, R. F., Brussee, J., and Ijzerman,A. P., Design-
ing active template molecules by combining computational de novo design and human
chemist’s expertise. J. Med. Chem. 2007, 50(8), 1925–1932.
66. Molecular Evoluator [computer software] CIDRUX Pharminformatics: Haarlem, The
Netherlands. www.cidrux.com
67. Douguet, D., Munier-Lehmann, H., Labesse, G., and Pochet, S., LEA3D: A computer-
aided ligand design for structure-based drug design. J. Med. Chem. 2005, 48(7),
2457–2468.
68. Schneider, G. and Fechner, U., Computer-based de novo design of drug-like molecules.
Nat. Rev. Drug Discov. 2005, 4(8), 649–663.
69. Mauser, H. and Guba, W., Recent developments in de novo design and scaffold hopping.
Curr. Opin. Drug Discov. Devel. 2008, 11(3), 365–374.
70. Voronkov, A. E., Baskin, I. I., Palyulin, V. A., and Zefirov, N. S., Molecular modeling of
modified peptides, potent inhibitors of the xWNT8 and hWNT8 proteins. J. Mol. Graph.
Model. 2008, 26(7), 1179–1187.
71. Nishibata, Y. and Itai, A., Confirmation of usefulness of a structure construction program
based on three-dimensional receptor structure for rational lead generation. J. Med. Chem.
1993, 36(20), 2921–2928.
72. Iwata, Y., Naito, S., Itai, A., and Miyamoto, S., Protein structure-based de novo
design and synthesis of aldose reductase inhibitors. Drug Des. Discov. 2001, 17(4),
349–510.
73. Heikkila, T., Thirumalairajan, S., Davies, M., Parsons, M. R., McConkey, A. G.,
Fishwick, C. W., and Johnson, A. P., The first de novo designed inhibitors of Plasmod-
ium falciparum dihydroorotate dehydrogenase. Bioorg. Med. Chem. Lett. 2006, 16(1),
88–92.
74. Ali, M. A., Bhogal, N., Findlay, J. B., and Fishwick, C. W., The first de novo-designed
antagonists of the human NK(2) receptor. J. Med. Chem. 2005, 48(18), 5655–5658.
75. Ji, H., Zhang, W., Zhang, M., Kudo, M., Aoyama, Y., Yoshida, Y., Sheng, C., et
al., Structure-based de novo design, synthesis, and biological evaluation of non-azole
inhibitors specific for lanosterol 14alpha-demethylase of fungi. J. Med. Chem. 2003, 46(4),
474–485.
76. Grembecka, J., Sokalski, W. A., and Kafarski, P., Computer-aided design and activity pre-
diction of leucine aminopeptidase inhibitors. J. Comput. Aided. Mol. Des. 2000, 14(6),
531–544.
77. Grembecka, J., Mucha,A., Cierpicki, T., and Kafarski, P., The most potent organophospho-
rus inhibitors of leucine aminopeptidase. Structure-based design, chemistry, and activity.
J. Med. Chem. 2003, 46(13), 2641–2655.
78. Alig, L., Alsenz, J., Andjelkovic, M., Bendels, S., Benardeau, A., Bleicher, K., Bourson,
A., et al., Benzodioxoles: Novel cannabinoid-1 receptor inverse agonists for the treatment
of obesity. J. Med. Chem. 2008, 51(7), 2115–2127.
Computer-Aided Molecular Design 315
79. Rogers-Evans, M., Alanine, A. I., Bleicher, K. H., Kube, D., and Schneider, G., Identifi-
cation of novel cannabinoid receptor ligands via evolutionary de novo design and rapid
parallel synthesis. QSAR Comb. Sci. 2004, 23(6), 426–430.
80. Stahl, M., Todorov, N. P., James, T., Mauser, H., Boehm, H. J., and Dean, P. M., A valida-
tion study on the practical use of automated de novo design. J. Comput. Aided Mol. Des.
2002, 16(7), 459–478.
81. Fechner, U. and Schneider, G., Flux (1): A virtual synthesis scheme for fragment-based
de novo design. J. Chem. Inf. Model. 2006, 46(2), 699–707.
82. Schuller, A., Suhartono, M., Fechner, U., Tanrikulu, Y., Breitung, S., Scheffer, U., Gobel,
M. W., and Schneider, G., The concept of template-based de novo design from drug-
derived molecular fragments and its application to TAR RNA. J. Comput. Aided Mol. Des.
2008, 22(2), 59–68.
11
Reaction Network
Generation
Jean-Loup Faulon and Pablo Carbonell
CONTENTS
11.1 Introduction ...................................................................317
11.2 The Challenges of Generating Networks ....................................318
11.3 Representation of Chemical Reactions ......................................319
11.3.1 Representation Based on Reaction Centers ........................320
11.3.2 Bond–Electron Matrices ............................................321
11.3.3 Representation Based on Fingerprints .............................322
11.4 Network Kinetics .............................................................324
11.4.1 Estimating Reaction Rates ..........................................324
11.4.2 Simulating Network Kinetics.......................................326
11.5 Reaction Network Generation Algorithm....................................328
11.6 Reaction Network Sampling Algorithm .....................................333
11.6.1 Concentration-Sampling Network-Generator Algorithm..........333
11.6.2 MC-Sampling-Network-Generator Algorithm.....................336
11.6.3 SMS Algorithm......................................................337
11.7 Concluding Remarks .........................................................339
References ............................................................................339
11.1 INTRODUCTION
Designing new drugs or chemicals, understanding the kinetics of combustion
and petroleum refining, studying the dynamics of metabolic networks, and using
metabolism to synthesize compounds in microorganisms are applications that require
us to generate reaction networks.As the networks may be quite large, there is a need to
automate the procedure. The purpose of this chapter is to present algorithms that have
been developed to handle that task. Section 11.2 introduces the problem and the chal-
lenges associated with network generation. Section 11.3 gives various methods used
to represent chemical reactions and reaction networks. Section 11.4 introduces tech-
niques to simulate the dynamics of reaction networks and, in particular, approaches to
estimate rate constants of the reactions. Section 11.5 presents deterministic techniques
to generate networks. It is not always possible to deterministically produce reac-
tion networks due to the combinatorial explosion of the number of species involved;
stochastic network generation methods are presented in Section 11.6.
317