Molecular Simulation - What does my molecule look like?

The computer representation of compounds can certainly be chemically inspiring. David Bradley asks whether they come close to modelling what really goes on in the molecular world.

Long gone are the days of chemists sitting at a desk to draw their molecules with a pencil and chemical template, with its bond lines, hexagons, and cyclohexane chains and boats. And it's the rare student who works with plastic ball and stick models. These days there are so many packages around that can render a chemical sketch in three dimensions onscreen within a few seconds, or import a fully-formatted structure from a database. Examples include CambridgeSoft's ChemFinder or a website like Ron Rinehart's Molecular Heaven which links to numerous resources for finding all kinds of molecular structures. With so many diverse sources and programs to display and manipulate compounds, there now seems to be little need to consider the molecules themselves too deeply.

The ArgusLab Molecular Modelling Program (now at version 3.0 beta), like many other programs, features extensive support for rendering molecular surfaces. It can build three-dimensional interactive molecules and, crucially, optimise structures for the entire periodic table so that it provides a conformation as close to the likely 'real' shape of a molecule in the real world. Computational techniques built into this and other packages, such as those from ACD's ACD/HNMR, allow aspects of chemistry such as nuclear magnetic resonance spectra to be predicted from the geometry-optimised structure.

Two dimensions
The early chemistry drawing packages used that most fundamental of structural representations - the 2D structure. This was akin to the paper-bound drawings made by an earlier generation using a template or even freehand, and provided chemists with a way to neaten up and automate the production of chemical illustrations for their research papers and presentations. The familiar drawing packages, ChemDraw, MDL Information Systems' ISIS/Draw, and ChemWindow, now at version 6, have all evolved tremendously over the years. Now, the latest version of a package, such as ChemDraw Ultra 6.0, will recognise the stereochemistry inherent in a molecule, check the valencies of every atom, clean up the bond lengths and angles, and even produce a systematic name from the structure using IUPAC chemical nomenclature rules.

The 2D structures and their ball-and-stick relatives hinge on the rules of covalent bonds described by G.N. Lewis. A single line represents a bond between the atoms at each of its ends. While chemists can 'see' very clearly what a 2D diagram represents, such a diagram encapsulates only one aspect of a chemical, the purported interconnectivity of its atoms at the covalent level. But we have to assume that molecules are at least three-dimensional objects given that we do not live in Flatland.

Rendering a 2D structure as a 3D molecule is the forte of a whole range of programs including ChemWindow (the parent suite of ChemDraw); the Chime program from MDL, which is web-browser based; and one of the original viewers, RasMol. Molecules-3D, from Molecular Arts Corporation, also provides fine structures. None of these are to be criticised; they are all powerful tools in capable hands. Accelrys' Materials Studio provides great control of the display of molecular structure. There is even the option of applying different coloured lighting effects to a structure, which might seem to be nothing but a gimmick on first glance, but can illuminate aspects of a structure not seen with a flatter 3D view.

No more leggy bonds
Display of molecule as 3D structures is amenable to all kinds of representations, from the basic 3D version of the simple stick diagram or the ball-and-stick, to the space-filling model. This latter is also known as a CPK model after the inventors of the concept, Elias J. Corey, Linus Pauling and W.L. Kulton. In the space-filling model, atoms are rendered as spheres with their scaled 'real' radius - their van der Waals radius, the distance out from the nucleus into which the electrons extend. This representation shows how the atoms may clump together as they might in the real world with no leggy bonds getting in the way. One additional chemical reality that is reintroduced with the space-filling model is the fact that just how bulky some chemical groups can be becomes immediately apparent as atoms jostle for position in a molecule.

A surfeit of surfaces
There is much more to a molecule than its atoms and bonds, though: there are the interactions between different parts of the molecule; and how they behave in solvents, when other molecules are present, or when they crystallise. Some of this behaviour can also be represented visually. While you are browsing molecules embedded in web pages, a program like ArgusLab can be used to show the electrostatic outline, or surface, of a molecule. This rendering provides a view of how a molecule 'feels' to solvent or other molecules. In one sense it smoothes out the troughs between atoms, but in a more scientific manner it represents the electrostatic potential of the molecule as a whole, with its highs and lows smoothed out between atoms, as there are no gaps in this field.

Proteins
When it comes to rendering in 3D a macromolecule - a nucleic acid, say, or a protein - then ball and stick representations, which were fine for simple small molecules, just look like so many entangled bead necklaces, while space-filling models engulf the interesting parts of the structure in a seemingly amorphous mass.

The protein scientists therefore have come up with their own 3D library of motifs and representations, which are all now incorporated into the drawing and manipulation software. A ribbon-like curl, a spaghetti string, and a flash of blue tape represent regions of secondary structure, the alpha helices, the long, plain stretches of amino acids, the beta-pleated sheets. Some of the packages allow the user to vary the parameters used to generate the protein model. For instance, they can be rendered as a flat ribbon, a shadowed 3D ribbon, or as a pure schematic where helices might be represented by chunky red cylinders and the beta-sheets as stark, pointing arrows.

Crystal structures, too, give little away of their internal symmetries when rendered as space-filling models. The polyhedral representation of co-ordination possible with Materials Studio, the aptly named Diamond and other packages, is much more adept at providing chemists with a clear view of their material. It allows one to gloss over the details of atom shape and view the overall crystal stacking and packing of a compound. Usually, the only parameter that limits the display is the amount of RAM installed on your computer.

It is possible to generate some rather beautiful three-dimensional views of crystal structures. If you have access to the Cambridge Crystallographic Database, for instance, you can import the data into Cambridge Soft's Chem3D, or any of a half dozen other packages. The CCDC has just added its 250,000th data set for a compound: isoindolo(2,1-a) pyrrolidino(2,1- c)(1,4) benzodiazepine-1,12-dione-6-ol. It also provides its own manipulation tools for handling the data in many different ways. Once imported into a 3D editing program, the compound can be tweaked and finetuned, then chemical groups adjusted or changed to display a derivative. A 2D structure program is often easier to work with in terms of swapping groups and correcting bond valencies and the like. Finally, the structure might be saved in one of several formats such as the mol file format for viewing in the Mercury crystal viewer from Cambridge or the MolView program from Jean-Michel Cense. The options are almost limitless in terms of display. Add a stereo display and you can get a feel for the genuine three-dimensionality of the compound too.

There is a 'but'
While chemists were restricted to print, molecules were plain. The advent of the computer and the increasing sophistication of visualisation tools have led to a revolution in understanding molecular structure. Chemists can now almost handle the molecules they discover and synthesise. Indeed, the likes of VRML (virtual reality mark-up language) provides the means to manipulate molecules in a seemingly very realistic way. Molecules can be pushed together, substrates docked with enzymes, allergens fed to antibodies and much more. The chemical content of molecules held in the very latest tools is far higher than is possible with a pen and paper; bonds, molecular orbitals, noncovalent interactions, surfaces, all kinds of additional information can be displayed. Individual atoms and chemical groups can be hyperlinked to related information, atomic co-ordinates, spectrographic data, and other physical parameters. Chemists have never had it so good.

None of these representations of molecules comes anywhere near what exists in reality. They are as accurate in the mind as one cares to see them. Some are as exacting in their detail as a Velasquez, but they all paint a picture of symbol and metaphor; others are Monet in their impression of those atoms and bonds; while the primary colours hint at Mondrian; other representations are as intelligible to the outsider as a Picasso.

They are fine art. But, do they answer the question: 'What does a molecule look like?' Only as well as art answers the question: 'What can you see?'

A molecular structure is a word, or perhaps a paragraph, in the language of chemistry; in what language it is spoken is a matter of taste and environment. The word is not the molecule itself any more than you can sit on the word 'chair'. What is certain is that chemists could not do without molecular visualisations just as the rest of us cannot do without words.