Biomedical Engineering Reference
In-Depth Information
CHAPTER 23
Distributed Workflows in
Bioinformatics
ARUN KRISHNAN
23.1
INTRODUCTION
The development of improved DNA (deoxyribonucleic acid) sequencing technologies
in the 1980s and 1990s heralded the start of a new era in biology. Biology and,
in fact, biomedical science have slowly transformed into a multidisciplinary arena
where there is a confluence of such diverse fields of research as genetics, molec-
ular biology, computer sciences, mathematics, biostatistics, and bioinformatics. A
vast amount of data is also being generated from DNA microarrays, mass spectro-
metry, DNA sequencing, and structure analysis. To be useful, the data acquired
with these technologies need to be processed and interpreted. This requires new and
innovative computational paradigms, techniques, and algorithms that can effectively
process the large volume of data. In addition, it requires large parallel computing
resources.
The field of high performance computing has undergone even more rapid and
drastic changes over the past few years. Improvements in the performance of proces-
sors and networks have made it feasible to treat collections of workstations, servers,
clusters, and supercomputers as integrated computing resources or
Grids
. However,
heterogeneity, while allowing for scalability on one hand, makes application devel-
opment and deployment for such an environment extremely difficult on the other. For
grids to be widely useful, application design must be flexible enough to exploit widely
differing and geographically distributed resources, matching application requirements
and characteristics with grid resources.
Grid computing, as a distributed computing framework, offers a powerful, high per-
formance computing environment, particularly for coarse-grained, data-parallel appli-
cations. The vision of geographically distributed, heterogeneous high-performance
systems [22] is quickly becoming a reality, especially with grid middleware tech-
nologies, such as Globus, rapidly maturing and reaching stability. Many large-scale
Search WWH ::
Custom Search