High Performance Computing for Computational Biology - Session Introduction

Computational biology has become a fully multidisciplinary eld, including components of mathematics, biology, chemistry, and computer science. Computational biology is essentially the computer-aided analysis of the biology of organisms. Since even a single genome or proteome contains an immense quantity of data, performing even a simple analysis on genome-scale data quickly turns into a computationally diÆcult problem. Thus, computational biology now requires high-performance computing and its related components in database systems, visualization, and computer engineering. The focus of this session is computational biology's growing need for highperformance computing. The rst purely computational steps of the human genome project have required vast amounts of computing equipment with, for example, large processor farms being used in both the private and public assemblies of the human genome. To the over-optimistic researcher, the simplest way to gain a factor of 10 performance increase is to use 10 processors. In practice, many critical issues degrade performance. Communication between the processing elements takes additional time, as does partitioning the problem and recombining the answers. Some algorithms cannot even be e ectively partitioned. As seen in some of the papers in this session, often the most signi cant overhead is the additional time required to adapt an algorithm to a high-performance system. The session received 26 papers, 11 of which the conference chairs selected