Biomedical Engineering Reference
In-Depth Information
Molecular epidemiology of infl uenza virus strains provides scientists with
clues about the temporal and geographic evolution of the virus. Researchers
from France and Vietnam are developing a global surveillance network based
on grid technology: The goal is to federate infl uenza data servers and auto-
matically deploy molecular epidemiology studies [26]. A fi rst prototype based
on AMGA [12] and the WISDOM production environment [26] extracts daily
from the National Center for Biotechnology Information (NCBI) infl uenza
H1N1 sequence data which are processed through a phylogenetic analysis
pipeline deployed on EGEE [3] and AuverGrid (http://www.auvergrid.fr)
e-infrastructures. The analysis results are displayed on a Web portal (http://g-
info.healthgrid.org) for epidemiologists to monitor H1N1 pandemics.
15.6
PERSPECTIVES
15.6.1
Introduction
IT technology is constantly evolving and new concepts have been emerging
in recent years. The new popular concept heavily promoted by the largest
IT companies is cloud computing. Integrating private or public clouds on e-
infrastructures would enrich the services offered to their customers. However,
a number of questions are still open related to the interoperability of the grid
middleware and cloud services, to the business model of the private clouds,
and to the security framework required for their integration.
Another very promising approach to increase the computing resources
available to the scientifi c community is to use graphical processors.
In this chapter, we will present activities we are currently developing on
graphical processors applied to life sciences. We will also discuss how all the
developments we have made for eight years are now converging toward mul-
tiscale modeling for system radiobiology.
15.6.2
Graphical Processors
General-purpose graphical processing units (GP-GPUs) were designed to
process more than the regular computer graphics, but while the classical CPU
computation performance evolution recently began to slow down, the GP-GPU
has continued to provide very signifi cant speedup. Ten years ago, developers
of high-performance computing applications started to port scientifi c software
from CPU to GP-GPU to make the most of it [27]. While a CPU possesses
few cores, each of them allowing the execution of one thread at a time, a
GP-GPU possesses a small number of streaming multiprocessors, each of them
allowing the parallel execution of numerous threads, supporting vector com-
puting in a SIMD (single instruction multiple data) approach. After the initial
success, GPU manufacturers started to work on friendlier application pro-
gramming interfaces (APIs) for general-purpose computation, and one is
now able to develop directly in languages which are close variants of the C
Search WWH ::
Custom Search