2 Physical Methods and Techniques Part (ii) Computer Graphics (Computer Aids to 0rga n ic C h em istry) By D. E. JACKSON Department of Pharmaceutical Sciences University of Nottingham Nottingham NG7 2RD 1 Introduction The explosive growth in computer technology over the past decade which allows modern computer systems to communicate rapidly over long distances to handle and store large amounts of information and to carry out calculations with ever increasing speed has revolutionized data acquisition and management procedures. A prerequisite of any system which is to be of benefit to users other than the computer specialist is an easy-to-use (user-friendly) interface between operator and machine. As the graphical capabilities of hardware systems continue to develop coupled with the chemist’s traditional use of pictorial representations to describe molecules there has been an increasing use of computer graphics to provide just such an interface for applications in chemistry.Graphically orientated programs are now used for various tasks from the storage and retrieval of information through to the display of molecular properties derived from the more traditional computa- tional chemistry packages. Evolving from these traditional computational techniques are new molecular simulation methodologies for the investigation and representation of molecular behaviour which are at least beginning to address the long-standing questions of molecular recognition and co-operative molecular actions. The increasing importance of these computational and graphical techniques has been reflected in both the formation of a Molecular Graphics Society and the introduction of specialist journals including the Journal of Molecular Graphics and more recently the Journal ofcomputer-Aided Molecular Design.A further recognition of their growing importance to the organic chemist is the inclusion of ‘computer graphics’ in Annual Reports for the first time. However the rather diverse nature of the topic does mean that an initial Report of this nature must be somewhat selective in the areas covered and cannot be regarded as a comphrehensive survey of the literature. What this Report will endeavour to do given those provisos is give the organic chemist who is not a specialist in this area an indication of recent developments in graphically orientated applications and highlight the areas in which such techniques may contribute.Selected references from earlier work will be quoted where relevant. 17 D. E. Jackson 2 Molecular Modelling Drug Design.-The steadily increasing interest in computer-aided molecular modelling (CAMM) over the past several years has been due in the main to the application of such techniques in the area of drug design. In response to the consequent upsurge in reported applications involving CAMM the Provisional Section Committee on Medicinal Chemistry of IUPAC' is currently publishing a survey on the availability of both hardware and software appropriate for CAMM together with guidelines on the standards to be adopted for publications reporting computer modelling in medicinal chemistry.2 This latter step more than any other probably represents the 'coming of age' of CAMM as an investigative tool in drug development.A cautionary note may be appropriate here however as a recent perspective by Dearing3 on the development of CAMM as an everyday tool for the non-expert has highlighted the problem that many of the methodologies may not yet be sufficiently well developed and robust that they can be applied without error by those who may not be experts in the field. Several articles and reviews covering the techniques and methodologies4 involving computer graphics and computational chemistry which can be applied to the study of molecular recognition' and drug action6 have appeared.A recent article by Marshall7 on computer-aided drug design divided the approaches into those where the active site of the drug is known those where it can be constructed by homology with known structures or predicted from the known amino acid sequence and finally those where the active site must be inferred from a comparison of properties seen in a series of ligands. Many such examples of computer-aided contributions to molecular recognition and drug design involving known active sites,8 predicted active sites and ligand comparisons" have been reported over the past year. The identification of quantitative structure-activity relationships (QSAR) in a series of compounds has long been an important computational tool for the develop- ment and optimization of biologically active materials.In recent years the incorpor- ation of three- and four-dimensional numerical data into QSAR studies has resulted from the introduction of graphical techniques in QSAR A new ' J. G. Topliss J. Med. Chem. 1988 31 2229. P. Gund D. C. Barry J. M. Blaney and N. C. Cohen J. Med. Chem. 1988 31 2230. A. Dearing J. Cornput.-Aided Mol. Design 1988 2 179. (a) 'Topics in Pharmacology Vol. 3' ed. A. S. V. Burgen G. C. K. Roberts and M. S. Tute Elsevier Amsterdam 1986; (6) A. E. Howard and P. A. Kollman J. Med. Chem. 1988,31 1669; (c) W. F. van Gunsteren Protein Engineering 1988 2 5. P. Zielenkiewicz and A. Rabczenko Biophys. Chem. 1988 29 219. (a) P. J. Goodford J. Med. Chem. 1984 27 557; (6) J.G. Vinter Chem. Brit. 1985 21 33; (c) C. H. Hassall ibid. 1985 21 39; (d) A. Krohn in 'Second SCI-RSC Medicinal Chemistry Symposium' ed. J. C. Emmett Royal Society of Chemistry London 1984 p. 109; (e) A. J. Hopfinger J. Med. Chem 1985,28 1133; (g) W. Ripka New Scientist 1988 54. ' G. R. Marshall Annu. Rev. Pharmacol. Toxicol. 1987 27 193. (a) H.A. Schreuder W. G. J. Hol and J. Drenth J. Biol. Chem. 1988 263 3131; (b) S. Neidle L. H. Pearl P. Herzyk and H. M. Berman Nucleic Acids Res. 1988 16 8999; (c) .C. Mukhopadhyay and V. S. R. Rao Znt. J. Biol. Macromol. 1988 10 217. (a)M.-J. Santoni C. Goridis and J. C. Fontecilla-Camps J. Neuroscience Res. 1988,20 304; (b)W. V. Williams H. R. Guy D. H. Rubin F.Robey J. N. Meyers T. Kieber-Emmons D.B. Weiner and M. 1. Greene Proc. Nutl. Acad. Sci. USA 1988 85 6488. (a) B. V. Cheney J. Med. Chem. 1988 31 521; (b) D. Mayer C. B. Naylor I. Motoc and G. R. Marshall J. Cornput.-Aided Mol. Design 1987 1 3. " A. K. Ghose and G. M. Crippen J. Med. Chem. 1985 28 333. Physical Methods and Techniques -Part (ii) Computer Graphics 19 approach using the technique of comparative molecular field analysis (CoMFA) has been described recently.12 This involves sampling steric and electrostatic fields surrounding a set of ligands in a similar manner to the probe interaction grids described earlier by G~odford,’~ and using molecular field comparisons to quantify structure and biological activity. A related microcomputer-based technique involving the generation of a hypothetical active site lattice (HASL) has been described recently by D0~eyko.l~ In this approach four-dimensional molecular lattices derived from Cartesian co-ordinates and a physicochemical descriptor such as hydrophobic- ity or electron density are used for the quantitative comparison of molecules.The merging of information from selected structures gives a composite lattice of points (the HASL) which may reflect the shape and binding properties of an active site. Molecular Mechanics.-At the foundation of any molecular modelling package is a procedure for the calculation of molecular energies. Currently only those tech- niques involving the use of molecular mechanics offer a sufficiently quick method to satisfy the requirements of the interactive molecular modeller.In principle molecular mechanics assumes that the energy of a system can be described by the sum of the energies of various mechanical (stretching bending torsional and van der Waals) and electrical terms such that Etotal = Estr + Ebend + Etor + EvdW + Ecoulombic (1) The exact mathematical nature of the individual terms together with the parameteriz- ation of the atom types constitutes the ‘force field’.’’ At present no force field exists which is capable of dealing with all atom combinations; therefore careful consider- ation must be given as to the accuracy of a given force field for a given situation. Clearly the development of accurate forms of the potential energy function to simulate wide-ranging molecular behaviour is crucial to the advancement of modelling techniques.To this end the Biosym Consortium has pooled academic and industrial resources for the development of more accurate force field parameters. As part of this research effort Hagler has recently described a method for determining force constants and optimal forms of the energy function for small molecules by the use of ab initio molecular energy surfaces.16 At present there are several commonly used force fields which have been described in the literature. These include CHARMM,17 MM218 and more recently MM3,19 AMBER,20 and the Consistent Valence Force Field (CVFF) used in Discover.21 The I* R. D. Cramer 111 D. E. Patterson and J. D. Bunce J. Am. Chem. Soc. 1988 110 5959. l3 P. J. Goodford J. Med. Chem.1985 28 849. 14 A. M. Doweyko. 1. Med. Chem. 1988 31. 1396. l5 (a) ‘Molecular Mechanics’ ACS Monograph Series 177 ed. U. Burkett and N. L. Allinger American Chemical Society Washington DC 1982; (6) D. B. Boyd and K. B. Lipkowitz J. Chem. Educ. 1982 59. 269. 16 J. R. Maple U. Dinur and A. T. Hagler Proc. Natl. Acad. Sci. USA 1988 85 5350. (a) B. R. Brooks R. E. Bruccoleri B. D. Olafson D. J. States S. Swaminathan and M. Karplus J. Comput. Chem. 1983 4 187; (6) L. Nilsson and M. Karplus ibid. 1986 7 591. la J. T. Sprague J. C. Tai Y. Yuh and N. L. Allinger J. Comput. Chem. 1987 8 581. 19 N. L. Allinger and J. H. Lii J. Compur. Chem. 1987 8 1146. 20 (a) S. J. Weiner P. A. Kollman D. A. Case U. C. Singh C. Ghio G. Alagona S.Profeta and P. Weiner J. Am. Chem Soc. 1984 106 765; (6) S. J. Weiner P. A. Kollman D. T. Nguyen and D. A. Case J. Compur. Chem. 1986 7 230. 21 (a) A. T. Hagler P. Dauber and S. Lifson J. Am. Chem. SOC 1983 101 5131; (6) ‘Discover Users Manual V2.4’ Biosym Technologies Inc. 10065 Barnes Canyon Rd Suite A San Diego CA 92121. 20 D. E. Jackson force field used in the COSMIC molecular modelling package has recently been described.22 A modification to the parameter set of CHARMM has been reported23 for the simulation of carbohydrate pyranose rings and its performance compared with the most developed carbohydrate potential energy surface PEF 422.24 The OPLS (optimized potentials for liquid simulations) potential function for proteins in which inter- and intramolecular non-bonded interactions expressed through coulombic and Lennard-Jones terms for peptide residues has been described.25 When combined with the bond stretch angle bend; and torsional terms from AMBER the AMBER/OPLS force field appears to be more accurate than AMBER alone in crystal simulations of cyclic peptides and the protein crambin although further validation in non-crystalline environments is necessary.One of the difficulties in molecular mechanics is its application to systems containing delocalized electrons; this is because of the need to determine a specific parameter set for each delocalized bond in the system. A method developed by Allinger uses an SCF calculation on the 7-electron system to determine bond orders of the conjugated system.From the relationship between bond orders and force field parameters it is possible to deduce parameters for specific bonds as they occur. Such techniques for dealing with conjugated hydrocarbons and ketones have been included in MM2( 82) with an extension to include conjugated nitrogen heterocycles in MM2(85),26 although limitations with the latter application have been recognized. Molecular Dynamics (MD).-In recent years attention has focused on the dynamic aspects of macromolecular structure and function. A combination of experimental X-ray and neutron diffraction crystallography together with a number of spectro- scopic techniques has increased our understanding of the dynamic behaviour of such systems. Alongside such experimental procedures theoretical studies involving molecular dynamic simulation methods have aided our understanding of macro-molecular atomic motions.Using an analytical potential of the type described above to express the energy of a system the negative derivative of the potential with respect to the co-ordinate gives the force on each atom F = ma = -dV/dr or -dV/dr = m d2r/dt2 where rn = mass of atom a = acceleration r = Cartesian co-ordinates of atom i. Thus by solving Newton’s second law of motion for each degree of freedom for all atoms [see equation (2)] it is possible to compute a trajectory for each atom as a function of time (ca. s steps). Several reviews on the techniques and applica- tions of MD simulations in the study of macromolecules have appeared.27 Berendsen 22 (a) J.G. Vinter A. Davis and M. R. Saunders J. Cornput.-Aided Mol. Design. 1987 1 31; (b) R. J. Abraham and I. S. Haworth ibid. 1988 2 125. 23 S. N. Ha A. Giammona and M. Field Carbohydrate Rex 1988 180 207. 24 K. Rasmussen Acta Chem. Scand. Ser. A 1982 36 323. 25 W. L. Jorgensen and J. Tirado-Rives J. Am. Chem. Soc. 1988 110 1657. 26 J. C. Tai and N. L. Allinger J. Am. Chem. Soc. 1988 110 2050. 27 (a) P. Kollman and W. F. van Gunsteren in ‘Methods in Enzymology’ Vol. 154 ed. R. Wu and L. Grossman Academic Press New York 1987 p. 430; (b) M. Karplus A. T. Brunger R. Elber and J. Kuriyan Cold Spring Harbor Symp. 1987,52,381; (c) R. Bruccoleri M. Karplus and J. A. McCammon Biopolymers 1986 25 1767; (d) A.T. Hagler in ‘The Peptides’ Vol. 7 Academic Press New York 1985 p. 213. Physical Methods and Techniques -Part (ii) Computer Graphics 21 has recently summarized the methods available for MD simulation and its applica- tions to complex molecular systems.28 Conformational studies on several ~eptide~~ and nucleic acid3' structures have recently been described along with enzyme active site sir nu la ti on^.^^ The dynamic behaviour of molecules in the crystalline environ- ment is now the subject of investigation. Recent reports include the study of despentapeptide insulin32 and P-cy~lodextrin.~~ Molecular dynamic simulations have also been used to study conformational differences between molecules in different environments. a-Cyclodextrin has been studied in aqueous solution and in crystalline form.3' Studies on the cyclic peptide ~yclo-(Ala-Pro-D-Phe)~ in the isolated and crystalline states have shown the energy of the crystal conformer to be in the order of 8 kcal mol-' higher than an isolated theoretical minimum.35 Even more interesting is the observation that as a result of these conformational differences the methyl groups of the alanine residue rotate more freely in the crystal structure than in the isolated ~eptide.~~ This example in particular serves to highlight the greater awareness of the dynamic nature of molecules even in the crystalline state and further serves to demonstrate the fact that crystal conformations do not necessarily reflect solution conformations nor the bioactive conformation of a drug molecule.With the continuing advances in n.m.r. techniques solution conformations of peptides and small proteins can now be determined. By the application of two- dimensional NO! experiments in particular many approximate interproton dis- tances up to ca. 5 A can be determined.37 By the inclusion of another harmonic term in the potential energy function3* such that where En,m,r, constrains the interproton distances to the experimentally determined values it is possible to drive the molecular refinement using energy minimization and molecular dynamics in accordance with experimentally observed data. Several reports describing the use of restrained molecular dynamics have appeared recently. These include conformational studies on the lipopeptide myc~subtilin,~~ the poly- peptides secretin:' human growth hormone releasing factor:' and potato car-b~xypeptidase~~ inhibitor and a cyclic peptide somatostatin analogue.43 Refinements 28 H.J. C. Berendsen J. Cornput.-Aided Mol. Design 1988 2 217. 29 (a) R. S. Struthers J. Rivier and A. T. Hagler Annu. Rev. New York Acad. Sci. 1985 439 81; (6) J. R. Somoza and J. W. Brady Biopol-vmers 1988 27 939; (c) R. Rone N. Manesis M. Hassan M. Goodman A. T. Hagler D. H. Kitson and V. A. Roberts Tetrahedron 1988,44 895. 30 M. Hirshberg R. Sharon and J. L. Sussman J. Biomol. Stmct. Dynamics 1988 5 965. 31 (a) A. D. MacKerell jun. L. Nilsson R. Rigler and W. Saenger Biochemistry 1988,27,4547; (6) F. K. Brown and P. A. Kollman J. Mol.BioL 1987 198 533. 32 S. Yun-Yu Y. Ru-Huai and W. F. van Gunsteren J. Mol. Biol. 1988 200 571. 33 J. E. H. Koehler W. Saenger and W. F. van bunsteren Eur. Biophys. J. 1987 15 211. 34 J. E. H. Koehler W. Saenger and W. F. van Gunsteren J. Mol. Biol. 1988 203 241. 35 D. H. Kitson and A. T. Hagler Biochemistry 1988 27 5246. 36 D. H. Kitson and A. T. Hagler Biochemistry 1988 27 7176. 37 D. M. LeMaster L. E. Kay A. T. Brunger and J. H. Prestegard FEBS Lett. 1988 236 71. 38 H. Kessler C. Griesinger J. Lautz A. Muller W. F. van Gunsteren and H. J. C. Berendsen J. Am. Chem. Soc. 1988 110 3393. 39 M. Genest D. Marion A. Caille and M. Rak Eur. J. Biochem. 1987 169 389. 40 G. M. Clore M. Nilges A. Brunger and A. M. Gronenborn Eur. J. Biochem.1988 171 479. 41 A. T. Brunger G. M. Clore A. M. Gronenborn and M. Karplus Protein Engineering 1987 1 399. 42 G. M. Clore A. M. Gronenborn M. Nilges and C. A. Ryan Biochemistry 1987 26 8012. 43 H. Pepermans D. Tourwe G.van Binst R. Boelens R. M. Scheek W. F. van Gunsteren and R. Kaptein Biopolyrners 1988 27 323. 22 D. E. Jackson in the solution conformations of several nucleic acids have also been reported.44 A combined distance geometry (DG) and restrained MD procedure has been applied to the lac repressor DNA binding domain to produce three-dimensional structures that satisfied NOE distance constraint^.^' Difficulties in using n.m.r. data and restrained molecular dynamics may arise however when multiple conformations of the molecule generate an n.m.r.parameter set which cannot be satisfied by just one conformation of the molecule. This has been found to be the case in the cyclic decapeptide anatamanide where at least two solution conformations are required.46 A new dynamics simulation methodology based on the combination of quantum and molecular mechanics potentials has recently been described by Karpl~s.~’ This method has been implemented within the context of thermodynamic perturbation theory48 using a modification of the CHARMM program to include quantum mechanical contributions evaluated by the MOPAC program suite. Applications range from the dynamic simulation of SN2reactions in solution through to ligand binding and enzyme-mediated transformations. Finally on the subject of dynamic motion in biomacromolecules the low- frequency collective motion which is now believed to be present in such systems has recently been reviewed.49 Although modelling of such motion is not yet possible the identification of low-frequency collective macromolecular motion may herald significant developments in our understanding of aspects of molecular recognition processes which are poorly understood at present.Molecular Orbital Calculations.-The second method for the computation of molecular energies and properties involves molecular orbital techniques. Several of the modelling packages (e.g. CHEM-X COSMIC) offer graphical interfaces to various MO programs thereby making techniques which were once the provision of the theoretician alone available to the organic chemist.The most commonly used MO programs range from semi-empirical (e.g. CNINDO,” AMPAC”) through to ab initio (e.g. Gaussian5*) calculations. No further discussion on MO techniques will be included here. A perspective on CAMM and MO calculations has recently appeared53 and an excellent practical guide to the application of these programs aimed particularly at the non-specialist is available in A Handbook of Computational Chemistry.54 44 (a) C. S. Happ E. Happ G. M. Clore and A. M. Gronenborn FEBS Lett. 1988 236 62; (b) C. S. Happ E. Happ M. Nilges A. M.Gronenborn and G. M. Clore Biochemistry 1988 27 1735; (c) G. M. Clore H. Oschkinat L. W. McLaughlin F. Benseler C. S. Happ E. Happ and A. M. Gronenborn ibid. 1988 27 4185.45 J. de Vlieg R. M. Scheek W. F. van Gunsteren H. J. C. Berendsen R. Kaptein and J. Thomason Proteins 1988 3 209. 46 H. Kessler C. Griesinger J. Lautz A. Muller W. F. van Gunsteren and H. J. C. Berendsen J. Am. Chem. SOC. 1988 110 3393. 47 P. A. Bash M. J. Field and M. Karplus J. Am. Chem. SOC.,1987 109 8092. 48 (a) G. M. Torrey and J. P. Valleau Chem. Phys. Lett. 1974 28 578; (b) P. Kollman S. Rao F. Brown V. Daggett G. Seibel and U. C. Singh in ‘Protein Structure Folding and Design 2’ ed. D. L. Oxender Alan R. Liss Inc. New York 1987 p. 215. 49 K.-C. Chou Biophys. Chem. 1988 30,3. 50 J. A. Pople and D. L. Beveridge ‘Approximate Molecular Orbital Theory’ McGraw-Hill New York 1970. 51 Dewar Research Group QCPE Bull.1986 6 4 (AMPAC QCPE 506). 52 J. S. Binkley R. A. Whiteside K. Raghavachari R. Seeger D. J. DeFrees H. B. Schlegel M. J. Frisch J. A. Pople and L. R. Kahn ‘Gaussian82’ Carnegie-Mellon University Pittsburgh 1982. 53 P. von R.Schleyer J. Cornput.-Aided Mol. Design 1988 2 223. 54 T. Clark ‘A Handbook of Computational Chemistry’ Wiley New York 1985. Physical Methods and Techniques -Part (ii) Computer Graphics 23 Software.-A comprehensive review55 by Vinter in 1986 on modelling systems contains details on both the hardware and software available for applications in drug design X-ray crystallography electron density fitting and molecular dynamics. TOM,56 a subpackage of one such electron density fitting program FRODO has subsequently been described for protein-ligand docking with interactive energy minimization.Other software packages developed for the mainframe/minicomputer systems and not covered by the earlier review include MACROMODEL,57 for use with small organic molecules through to large biopolymers using MM2( 85) AMBER and the AMBER/OPLS force fields; CHEM-X,58 which offers a range of modules for modelling various molecular structures from small organic organometallic and inorganic through to macromolecular proteins; COSMIC and ASTRAL which represent comprehensive computational chemistry and display packages used primarily for investigating the properties of small organic molecules as an aid to drug design and based on the COSMIC force field; BRAGI,59 a protein modelling system with interface to the AMBER force field; QUANTA6’ (based on the CHARMM force field) a comprehensive open architecture modelling package for application to small organic molecules biological polymers and inorganic materials; MOLCAD,61 an interactive three-dimensional molecular display package with inter- faces to other program systems including AMPAC and GRID; MOL3D,62 a modular interactive graphics program for the display and conformational analysis of molecules containing up to 256 atoms; MANOSK,63 for the display manipulation and analysis of both small and large molecular systems; PEPCRE,@ for the interac- tive modelling of oligopeptides.The development in recent years of powerful desktop ‘personal computers’ has allowed the transfer of some molecular modelling techniques from mainframe/mini systems to PCs.A comparative review of molecular modelling software for the IBM PCs covering CRYS-X ALCHEMY CAMSEQ/M CAMSEQ/PC MGP MOL- GRAF and PROMODELER I has recently been written.65 Other software packages available for this system and which have been reviewed include ChemCad,66 an interactive graphics program which can be used to create input files for MM2 and the AMPAC suite of programs; a Molecular Mechanics Package (MS-DOS Com- puters):’ consisting of a structure input program an energy minimize program based on Allingers MM2 force field and a draw program for viewing the structures 55 J. G. Vinter in ‘Topics in Pharmacology Vol. 3’ ed. A. S. V. Burgen G. C. K. Roberts and M. S. Tute Elsevier Amsterdam 1986 p.15. 56 C. Cambillau and E. Horjales J. Mol. Graph. 1987 5 174. 57 C. Still ‘MacroModel Version 1.5’ Columbia University Department of Chemistry 507 Havemeyer Hall New York NY 10027 1986. 58 ‘Chem-X’ Chemical Design Ltd Unit 12 7 West Way Oxford. 59 D. Schomburg and J. Reichelt J. Mol. Graph. 1988 6 161. 60 ‘Quanta’ Polygen Corporation 200 Fifth Avenue Waltham Massachusetts 02254. 61 J. Brickmann ‘Molcad’ Technische Hochschule Darmstadt Petersenstrasse 20 D-6100 Darmstadt West Germany. 62 D. Pattou and B. Maigret J. Mol. Graph. 1988 6 112. 63 J. Cherfils M. C. Vaney I. Morize E. Surcouf N. Colloc’h and J. P. Mornon J. Mol. Graph. 1988 6 155. 64 C. W. van der Lieth J. Palm A. Sundin R. E. Carter and T. Liljefors J.Mol. Graph. 1987 5 119. 65 M. Sadek and S. Munro J. Cornput.-Aided Mol. Design 1988 2 81. 66 E. L. Clennan J. Am. Chem. SOC.,1987 109 2229. 67 M. M. Midland J. Am. Chem. SOC.,1986 108 5042. 24 D. E. Jackson and producing hard copy; Molecular Graphics on the IBM PC microcomputer,68 a version of which is also available for the Apple rnicro~omputer,~~ and pdViewer7’ are drawing and visualization packages for the three-dimensional representation of molecules; Visual Molecules and Molecular Parameters7’ generate graphical display from crystallographic data and allow output of tabulated molecular parameters such as bond lengths and bond angles. Other systems which are known to be available for the IBM PC include Desktop Molecular M~deller.~~ For the Apple Macintosh Chem3D is a limited molecular modelling display program which has been reviewed.73 A molecular graphics system (MGC) for the BBC microcomputer is also available74 in which a dedicated graphics display co-processor has been developed to enhance the graphics capabilities of the BBC.Limited molecular mechanics routines have been included to form the basis of a molecular modelling teaching aid. 3 Computer Aids to Synthesis Planning Several databases have been developed to handle literature reports of chemical reactions. Typically a graphical interface is used to construct a full structure or fragment specification for searching the database for reactions conditions and bibliographic entries. Examples of organic reaction databases include ORAC,75 SYNLIB,76 and REACCS.77 In addition to straightforward databases several ‘knowl- edge-based systems’ or ‘expert systems’ have been developed.Characteristically these combine a knowledge base with a number of concepts models and generalized rules such that by the application of identifiable transforms (disconnections) the program works retrospectively from product to possible starting materials. Examples of ‘expert systems’ include LHASA78 and SYNCHEM II.79 A number of programs have also been developed which implement a generalized set of rules governing reactivity for different classes of organic reactions in order to predict the outcome of a proposed step given starting materials and reaction conditions. Such systems which include EROS,” SYNGEN,81 and CAMEO have no reliance on any database and cannot therefore be used to search for literature data.A recent report on the development of CAMEO8* indicates that this program can now evaluate various reaction types including nucleophilic electrophilic peri- cyclic oxidative and reductive. A comprehensive coverage of reactions used in the synthesis of heterocyclic compounds has also been included. 68 M. A. Fox and D. Shultz J. Am. Chem. SOC.,1986 108 7882. 69 D. R. Dalton and G. R. Webster jun. J. Am. Chem. SOC.,1987 109 2862. 70 J. V. Paukstelis J. Am. Chem. SOC.,1988 110 4098. 71 C. J. Burrows J. Am. Chem. SOC.,1987 109 5056. 72 ‘Desktop Molecular Modeller’ Oxford Electronic Publishing Oxford University Press Walton Street Oxford.73 D. S. Allen J. Am. Chem. SOC.,1988 110 7261. 74 ‘Chemdata Molecular Graphics System’ Chemdata Ltd Wendron Helston Cornwall. 75 A. P. Johnson Chem. Brit. 1985 21 59. 76 J. Boother Chem. Brit. 1985 21 68. 77 ‘REACCS’ Molecular Design Ltd 2132 Farallon Drive San Leandro CA 94577 U.S.A. 78 E. J. Corey A. P. Johnson and A. K. Long J. Org. Chem. 1980,45 2051. 79 H. L. Gelernter A. F. Sanders D. L. Larsen K. K. Agarwal R. H. Boivie G. A. Spritzer and J. E. Searleman Science 1977 197 1041. 8o J. Gasteiger and C. Jochum Top. Curr. Chem. 1978 93. 81 J. B. Hendrickson and E. Braun-Keller J. Comput. Chem. 1980 1 323. 82 M. G. Bures and W. L. Jorgensen J. Org. Chem. 1988 53 2504. Physical Methods and Techniques -Part (ii) Computer Graphics CHIRON83 is an interactive program for the analysis and perception of stereochemical features in molecules and the selection of chiral precursors in organic synthesis.4 Other Chemical Databases For general chemical queries using a simple graphical interface to generate structures or sub-structures it is possible to search the chemical literature by using the CAS ONLINE system available through STN International or using the DARC system through Telesystemes Q~estel.~~ Both are based on the CAS Registry System which was developed in the 1960s to index chemical substances reported in the literature. A more recent addition to the on-line chemical information which is available is the Beilstein databa~e.’~ The first release in the last quarter of 1988 via STN International contains the Handbook heterocyclic compounds from volumes 17-27 of the Basic Series to the supplementary Series E IV inclusive.Software for use on local PCs for the creation of search files for this database has also been described.86 An alternative to the on-line database is a structure analysing program SANDRA87 (Structure and Reference Analyser Program for the Beilstein Handbook of Organic Chemistry) available for the IBM PC. This program allows a graphical input to query where the compound ought to be located in the 340 or so volumes of the Beilstein Handbook thereby alleviating the need to understand the specific indexing system used by Beilstein. Several MEDCHEM QSAR databases are under development at Claremont within the MEDCHEM Project at Pomona College.88 These include programs for the estimation of log P (CLOGP) and molar refractivity (CMR) and a generalized chemical information database (THOR).An expansicn of THOR into a database (MENTHOR)89 for the storage and retrieval of three-dimensional co-ordinate and charge information as well as the more traditional biological and physical properties has recently been described. The Cambridge crystallographic database” and the Brookhaven protein data bank” remain the major depositories for crystallographic data and are important sources of information to both crystallographers as well as molecular modellers. 83 S. Hanessian ‘CHIRON’ University of Montreal Department of Chemistry Quebec Canada.84 P. Rhodes Chem. Brit. 1985 21 53. 85 ‘Beilstein Brief No. 2’ Springer-Verlag Tiergartenstr. 17 Heidelberg FRG 1988. 86 ‘Molkick version l.O’ Springer-Verlag Dept. New Media/ Handbooks Tiergartenstr. 17 Heidelberg FRG 1988. 87 S. R. Heller J. Am. Chem. Soc. 1987 109 5055. 88 ‘MEDCHEM’ Medicinal Chemistry Project Pomona College Claremont CA U.S.A. 89 Y. C. Martin E. B. Danaher C. S. May and D. Weininger J. Cornput.-Aided Mol. Design 1988 2 15. 90 F. H. Allen S. H. Bellard M. D. Brice B. A. Cartwright A. Doubleway H. Higgs T. Hummelink B. G. Hummelink-Peters 0. Kennard W. D. S. Motherwell J. A. Rogers and D. G. Watson Acta Crystallogr. B 1979 35 2331. 91 F. C. Bernstein T. F. Koetzle G. J. B. Williams E. F. Meyer jun. M. D. Brice J. R. Rogers 0.Kennard T. Shimanouchi and M. Tasumi J. Mol. Biol. 1977 112 535.