Section 17.2.2. Background – the evolution of molecular graphics hardware and software

The complexity of molecular structure and the fact that these sub-microscopic objects of study are not directly visible have necessitated the use of physical or pictorial representations to aid in interpretation, manipulation and understanding. Illustrations and models made of wood, plastic or metal served these purposes from the development of the original theories of molecular structure through to the first nucleic acid and protein structures solved in the 1950s. Over the following years, computer graphics has evolved into a significant and ubiquitous technology, helping to sustain the explosive growth of macromolecular structure research. Today, computer graphics pervade the activities of much molecule-based research, from quantum chemistry to molecular biology.

Computer-based molecular graphics can be traced back to 1948 and the X-RAC project of R. Pepinsky at Pennsylvania State University (Pepinsky, 1952). Pepinsky developed an analogue computer to carry out the Fourier transformation of X-ray structure factors to produce electron-density maps. Integrated within X-RAC was an oscilloscope that could display the contours of the electron density (Fig. 17.2.2.1). These displays were, to my knowledge, the first computer-generated images of molecular structure. Crystallographers from around the world came to Pennsylvania State University to use X-RAC and to marvel at the speed and automation possible in the solution of molecular structures. While the digital revolution quickly overtook the analogue approach, X-RAC clearly set the precedent for molecular scientists as early implementors and adaptors of computational and graphics technology.

Figure 17.2.2.1| top | pdf | One bay of X-RAC showing coefficient panels and the display oscilloscope. Inset: photo from the oscilloscope, showing a region of the phthalocyanine Fourier map. Reproduced from Pepinsky (1952).

In the 1960s, two seminal projects laid the foundation for modern molecular graphics. Early in the decade, Johnson's ORTEP (Johnson, 1970) program became widely available, allowing crystallographers to produce illustrations of three-dimensional (3-D) molecular structures on a pen plotter. These black-and-white line drawings of ball-and-stick models were used both for working drawings during structure analysis and for creating illustrations for publication. A few years later, experiments lead by Levinthal under Project MAC (Levinthal, 1966) at MIT pioneered the interactive display and transformation of 3-D molecular structures on a computer screen. By the end of the decade, the groundwork for molecular graphics was set: ORTEP convincingly demonstrated to a large number of scientists that the computer could be used as an alternative to the human hand to produce accurate drawings and stereoscopic pairs for the analysis and communication of the results of structural research. Project MAC showed that the computer could be used as an interactive environment in which to model and simulate on the molecular scale. These two projects helped define the two broad functions of molecular graphics: publication graphics, for which clarity of presentation is the essential goal, and working graphics, for which rapid feedback and high interactivity are the key elements.

In the 1970s, 3-D interactive computer-graphics systems became commercially available. Hardware offerings from companies such as Evans and Sutherland, Vector General, and Adage prompted a number of laboratories to develop interactive molecular-modelling software. Several of these early systems were devoted to the task of building an atomic model of a protein into the crystallographically derived electron-density map. Programs such as BILDER (Diamond, 1982), MMSX (Barry & McAlister, 1982), FRODO (Jones, 1978) and GRIP (Wright, 1982) began to replace metal Kendrew models and the cumbersome `Richards Box' optical comparitors. This application, more than any other, sold these expensive ($100 000) monochrome line-drawing graphics devices to the molecular-research community. Moreover, during that time, biomolecular structure determination was a major civilian consumer of 3-D interactive graphics devices.

Technical, commercial and scientific advances in the 1980s prompted enormous growth in the use of molecular graphics. As late as 1983, a worldwide list of laboratories using high performance graphics computers for molecular work could be maintained – the number was below 100 (Olson, 1983). By the end of the decade, that number grew into the thousands, and utilization spread beyond any ability to track it. At the beginning of the decade the expensive vector-graphics terminals were the only way to achieve interactive 3-D display. Ten years later, the colour raster display had taken over the interactive computer-graphics market, driving prices down and broadening display capabilities from lines and dots to include shaded surface representation. Early in the 1980s, several academic software packages, such as GRAMPS/GRANNY (O'Donnell & Olson, 1981; Connolly & Olson, 1985), MIDAS (Ferrin et al., 1988) and HYDRA (Hubbard, 1986), went beyond electron-density fitting to provide general graphics functionality for examining molecular structure and properties. Over the decade, the remarkable evolution of computer hardware – the advent of microprocessors, very large scale integration (VLSI) devices, personal computers and scientific workstations – increased the accessibility of molecular graphics. By the mid-1980s, the demand was such that several commercial companies had been established to market molecular graphics and modelling software. By the end of the decade, structural scientists in academic and industrial research settings had a wide variety of use-tested hardware and software platforms with which to perform molecular modelling.

The 1990s witnessed remarkable advances in the technology and sociology of computing as well as in the science of molecular structure and design. Moore's law of the microcosm, which estimates that `the effectiveness of microprocessors doubles every 18 months', continued to track growth accurately. Thus the performance-to-cost ratio of late 1990s computers was a millionfold higher than those of the mid-1960s. A 1997 200 Nintendo-64 game machine was faster and had more memory and far superior graphics than the Control Data 6600 supercomputer and peripherals of the 1960s, and, with its optional ($70) disk, bettered almost every technical specification of a 1980 VAX 11/780. Gelder's law of the telecosm posited in 1993 that `bandwidth will treble every year for at least the next 25 years'. This, coupled with Metcalf's law, that `the total value of a network to its users grows as the square of the total number of users' implies that the `teleputer', or non-localized computing, is becoming the computational environment of the future.

While software development continues to lag behind hardware growth, the emergence of the World Wide Web and the concepts of network-based computing have catalysed a rethinking of the nature of software, its development, distribution and inter-operability. The concepts of cyberspace and `virtual reality' have been implanted into the minds and expectations of the general public, promoting a renaissance in user-interface exploration and development. It is transforming the computer from a window through which to look into a portal through which to step. Suddenly, the other senses – sound, touch, taste and smell – can become part of the computational experience.