How computing power is driving the advances in biotechnology

Munich 30 November 2000At the Sanger Centre in the United Kingdom, Bill Blake, the Vice President of Compaq worldwide for High Performance Technical Computing, gave an overview of this market and importance, and highlighted Compaq's role and investments in new Alpha processors as well as their road map. The speaker particularly focused the field of protein research.


From genes and the research agents, and diagnostic and therapy, one moved to proteins, which results in therapeutic proteins and antibodies. Bill Blake mentioned that there are 120 to 280 therapeutic proteins, but less than 5 percent have been developed. There exists a total number of 5000 to 10.000 novel drug targets, but currently the pharmaceutical industry is only based on 500 drug targets.

Mr. Blake sees the genomic industry in three waves: the early pioneers with Genetech, Amgen, Biogen; the industrialisation with human genome science, Incyte and Celera; and the industrialisation of functional genomics with new technologies and new companies. The revenue growth for the bioinformatics IT market is expected to grow from more than 500 million US$ in 2000 to more than 1 billion in 2001 and more than 2.2 billion in 2002.

This has the following reasons:

  • computational biology develops rapidly
  • sequence data doubles every 12 months (2 x computing, 4 x storage)
  • complex analysis routines produce more data
  • improvement in algorithms and hardware efficiencies

There are three different elements besides availability (24x7), scalability and upgradability, which influence the selection of the vendors. The computation requires fast processors, large memory, 64 bits and lots of processors for the assembly, sequence searching and annotation codes. The data storage needs very large disk space and storage area networks. There, reliable fast back-up and retrieval capabilities are necessary, a field in which Oracle is standard. The last issue is the data distribution and integration. Here, a good response time for accessing and sending the data within a Web-based technology is a must.

Bill Blake listed the advantages of a Compaq solution: the best delivered bio-application performance with 64-bit solutions, optimised applications, the sometimes dramatically reduced run time, and a top price/performance. Compaq is able to deliver pre-staged and tested complex systems. It offers StorageWorks for reliability and flexibility, and has planning, deployment, and management services.

Mr. Blake also mentioned a large computational bioinformatics benchmark example, the reasons why a major U.S. commercial genomics company chose Compaq. The other Unix vendor needed 11 days (264 hours) to run the job and 3 days and 15 hours (87 hours) in the optimised version. Compaq only needed 7 hours in the first stage and 5 minutes in the optimised programme.

According to Bill Blake's opinion, Compaq Alpha is the leading platform in bioinformatics, as for example the Sanger Centre, Washington University, St. Louis, the MIT Whitehead Institute, as well as the French Centre National de Séquençage utilise Alpha processors in their capacity as research institutes. In addition, Celera Genomics, Geneva Proteomics, Incyte Pharmaceuticals, and many more companies in industry have installed this platform.

Additionally, Mr. Blake described the actual and coming Alpha processors. His reason number 1 for choosing Compaq and Alpha is the processor. The current Alpha, EV67, has 667 MHz clock rate, leading to1.3 GFlop/s, the memory bandwidth is 2.4 GByte/s peak and 1.3 GByte/s measured. This is an 8-fold increase compared to the EV5. Moreover, the highly optimising compilers deliver the performance to the user.

The EV6 (21264) is scheduled to .5 to 1+ GHz, 1 to 2 GFlop/s. EV67 is the actual one, next is EV68 with 833 MHz, 1.67 GFlop/s. The EV7 (21364) has a faster clock, 1 to 1.5 GHz, 2 to 3 GFlop/s, large L2 cache on-chip. RAMBUS increases the bandwidth to 12.6 GByte/s, a four- to five-fold improvement. By the on-chip router, this processor is ready for glueless Shared Memory Processors (SMPs). The EV8 (21464) has a faster clock, 1.5 to 2 Ghz and is an 8-wide superscaler. It has a doubled instruction rate and delivers 6+ GFlop/s. It builds on EV7 structure. The simultaneous multi-threading boosts the sustained performance.

Compaq invests 300 million US$ per year for the Alpha and the operating system development. Mr. Blake also mentioned another aspect, namely the very fast and very scalable interconnect. The user has the choice between Myrinet and Quadrics interconnect. Last but not least, the system software is a differentiator. Some highlights to be considered constitute the parallel file system spanning multiple physical file systems, the coherent file access resource job management, Load Sharing Facility (LSF), and the parallel tools.

Uwe Harms

[Medical IT News][Calendar][Virtual Medical Worlds Community][News on Advanced IT]