Computer-aided drug design remains an embryonic field, but it is one that has enormous growth potential as the time taken to bring a drug to market stays constant and the costs of research and testing continue to soar. It is also just the kind of multi-disciplinary activity at which Britons, given the chance, excel. And one small British company announced last week that it had created an ingenious collection of applications that promise to take the process of creating new drugs by computer several steps forward. The announcement was also the first bit of really good news for many months for the thoroughly chastened flagship Norwegian computer manufacturer Norsk Data A/S, whose minicomputers were chosen ahead of machines from many of the more glamourous manufacturers by virtue of the fact that they are much easier to microcode so as to optimise performance of what is an extremely large and complex piece of software. Katy Ring found out all about it.
The small computer-aided drug design company Proteus Biotechnology Ltd has launched its flagship product the BioEngine through a joint marketing agreement with Norsk Data UK which provides the necessary hardware for the system (CI No 1,177). The system has been developed by Proteus, a company formed two years ago with funding and management provided by Imseco Medical Services. The company is composed of academics who have research facilities at the University of Manchester’s Institute of Science & Technology, the London School of Tropical Medicine, and INRA, Paris, as well as working in collaboration with the Strathclyde Unit for Drug Design. The minimum requirement for running the software is a daunting 32Mb of memory, and 2Gb on disk – well it does need for example to store all known DNA structures – and it is designed for use Norsk Data’s 5000 Series of super minicomputers, from which it can be linked to Norsk Data terminals or to IBM PS/2s at the front, to a Transputer bank for neural networking, to Silicon Graphics workstations for graphics and, via Fortran, to a supercomputer such as a Cray for back-end intensive arithmetic if it is required. Metabolic rate The BioEngine is described as a highly integrated system of software and hardware embodying theoretical methods and an expert system based on use of empirical data. It enables design to start at the research stage, so that whole new molecules can be created, which do not necessarily exist in nature. Consequently, it can design new drugs and chemicals based on chemical and bio technological products. For, although it is geared for biological applications, work with the chemical applications company Frazier Williams has led to a design that can work in a variety of disciplines. Thus, it can work with flexible molecules such as TRH which control the human metabolic rate, and create new variants to be used in the treatment of strokes. Because the BioEngine enables design to enter at an early research stage it can also screen the feasibility of projects. The pharmaceutical market at which the product is largely targeted is very fragmented with 30 major companies such as DuPont and ICI, and hundreds of smaller specialised bio-technological companies. Each new pharmaceutical product costs about UKP100m and 12 years to develop and complete trials. Proteus believes that the BioEngine can reduce the cost of drug development for small companies by up to 90% and cut two to three years time off the number of development years, thus enabling companies to reach the patent stage earlier. For example, Proteus, which itself is in the business of drug design has already registered a number of patents, including a synthetic peptide-based vaccine against AIDS. Although toxicity tests still have to be carried out for drugs designed on the BioEngine, the company claims that its product is green in that it reduces animal experimentation at the research stage. Another of the product’s features highlighted by Proteus was the fact that it is driven by Global, a fifth generation programming language and expert system. Machine learning This sy
stem enables the computer to learn from and with the scientist, making it a handy tool for companies short of chemical physicists, molecular biologists and so on, since they can store solutions on disk for the drug designer to use. Indeed, the BioEngine is being promoted as more of a concept than a fixed system because it is open-ended in design. Proteus approached Norsk Data for a joint marketing agreeement, rather than DEC (which is active in this market in partnership with the US company Polygen (CI No 629)) on the grounds that it is easier to change the microcode on the ND5000 Series than it is on VAXes, which means that the software can reach Cray speeds on Norsk Data boxes for some tasks. The start-up kit for a small laboratory or a company running a pilot project would probably be in the region of UKP25,000, which would include the software, a Norsk Data supermini from the ND5000 Series, the requisite communications materials as well as consultancy. The first such kit is going in to an unidentified European pharmaceutical company in June, and while the initial market for the kits is the pharmaceutical industry, the systems will also be targeted at the agro-chemical and veterinary science industries – an animal neutering agent is currently proving successful in animal trials. With the launch of the BioEngine both Norsk Data and Proteus are looking, in the long term, to a sizeable chunk of the designer drug market, a market in which IBM and DEC are also interested (CI No 1,177).
