The degree to which visualization can be supported by a generic workstation is a function of the workstation's overall system performance, specific hardware characteristics, the standards supported by the system, and their affect on overall performance, including the availability of software drivers and other interface technologies that may affect researcher's use or system performance.
Visualization is one of the most computationally intensive processing challenges for workstation- based applications. Dedicated rendering workstations typically have the multiple, highest-speed processors commercially available, a gigabyte or more of RAM, and at least one fast hard drive for maximum image throughput. Special hardware or firmware graphics accelerators are typically used to handle operations such as providing perspective, shading of figures, and refreshing the screen quickly between redraws of images being manipulated in hardware that would otherwise have to be communicated through the computer bus, executed on the main CPU, and then communicated to the video card.
Software also profoundly affects the range of possibilities for visualizing simple graphics or complex 3D models. A visualization package written in Assembler by an accomplished programmer may outperform a poorly written application even when the later is running on higher-performance hardware. For example, if the application doesn't recognize the hardware graphics accelerator or make up for the maximum amount of RAM available, the visualization application won't deliver as much performance as possible. Similarly, the operating system may enhance or limit the
performance of the visualization application. Support for OpenGL-compatible graphics accelerators and multiple processors is a basic requirement. Compatibility with industry-standard drivers, storage media, and networking protocols is also essential.
When it comes to visualization on generic hardware, standards are a double-edged sword. The high performance that is possible with specific graphics accelerators and software drivers and applications may make the workstation virtually incompatible with any other applications, even other visualization applications. The creation of a dedicated, high-speed visualization workstation by adding coprocessor boards and special drivers on an otherwise generic workstation may be a viable option if the
enhanced performance and time savings is worth the additional expense and resources required to maintain an additional system.
On the Horizon
Virtual reality—the use of computers to immerse the user in a multimedia environment that's rich enough in synthetic cues to make the simulated environment seem real—has great potential in bioinformatics R&D. In the general marketplace, the commercial uses of virtual reality technology include virtual prototyping, museum displays, design evaluation, architecture, trade show displays, engineering, aerospace simulation, collaborative engineering, game development, and education. Most of these applications of the technology translate directly to bioinformatics applications. For example, the virtual prototyping of the functionality of running shoes or tractors isn't conceptually different from prototyping drugs and their effects on different human protein binding sites. Just as many of the traditional museums have been placed online to allow access to those who don't have museums in their communities, so virtual tours of protein molecules allow researchers and students access to data in a form that they couldn't otherwise access.
Design evaluation, which involves illustrating how a device or apparatus will look, can also be applied to protein structures. Virtual reality visualization methods can illustrate, for example, the different shapes that a protein molecule might assume with changes in local pH or temperature. Similarly, just as virtual architecture applications allow potential clients to experience the finished product before it's built, a virtual reality model of a protein structure allows researchers to work with 3D images of molecules before they're actually synthesized. The advantage of this approach is that it allows potential problems to be identified before resources are invested in developing the molecule. To date, the greatest commercial use of virtual reality in molecular biology is in the form of booth attractions at trade shows. The pharmaceutical industry spends several hundred-million dollars annually on the marketing of drugs at major medical conferences, and virtual reality and other forms of visualization technology are commonly used to attract future prescribers to their booths and to quickly communicate the mechanism of action and relative efficacy of their drugs.
Similarly, in the aerospace industry, the practical application of virtual reality includes everything from turbine design to flight simulation training for pilots and support personnel. Much of this is in the form of collaborative engineering, where engineers share models and interact online.
Collaborative engineering has been used for years in the automotive and aerospace industries to design subsystems and test their functionality before actually creating them. The result is that ineffective designs are disposed of before they make it to the prototyping stage, saving the companies time and money.
Closely related to virtual reality entertainment systems in which combatants donning virtual reality helmets immerse themselves in battle situations is the use of virtual reality in education. Several medical boards have invested heavily in virtual patient encounter systems in which physicians interact with animated, talking 3D patient simulations. These virtual reality systems allow medical students, residents, and physicians to develop their clinical pattern-recognition skills before interacting with patients suffering from the conditions being studied.
These and other applications of virtual reality have obvious application in molecular biology and bioinformatics research. For example, in the area of education, there is a significant gulf in what traditionally educated health care professionals and researchers understand about the bioinformatics arena. Similarly, virtual reality technologies can be used to enable students, researchers, and
Endnote
Visualization is one of the most active areas of R&D in bioinformatics. One reason that visualization technology is so advanced today is the huge investment over the past several decades in the area by the military establishment. Consider that the development of the first bitmapped screens were
supported by the military because the screens could track the trajectory of missiles more precisely than a simple grid of "Xs" on a character-oriented screen. Another reason for the rapid advances in the field is the parallel work in visualization being conducted in fields as diverse as the military, medicine, and weather forecasting. For example, based on the interfaces developed for use in clinical medicine, such as fMRI, the next generation of user interfaces used in bioinformatics will likely inherit some of this higher-level biological focus.
Bioinformatics visualization requirements, especially those related to 3D rendering of protein
structures and modeling protein-protein interactions in real time, will certainly drive development in high-end computing, including supercomputer and grid computing. The challenge for the
bioinformatics community is to devise visualization techniques and related technologies that are easily shared, capable of being supported in the long-term, and ones that provide developers of next- generation hardware and software with a viable target to support.