Molecular Structure Determination on a Computational and Data Grid - Parallel Computing for Bioinformatics and Computational Biology - page 583

CHAPTER 24
Molecular Structure Determination
on a Computational and Data Grid
RUSS MILLER and MARK L. GREEN
The focus of this chapter is on the design and implementation of a critical com-
puter program in structural biology on two computational and data grids. The first is
the Buffalo-based Advanced Computational Data Center (ACDC) Grid, which uses
facilities at SUNY-Buffalo and several research institutions in the greater Buffalo
area. The second is Grid2003, an international grid established late in 2003 primar-
ily for physics and astronomy applications. We present an overview of the ACDC
Grid and Grid2003, focusing on the implementation of several new tools that we
have developed for the integration of computational and data grids, lightweight job
monitoring, predictive scheduling, and opportunities for improved Grid utilization
through an elegant backfill facility. A new computational framework is developed
for the evolutionary determination and an efficient implementation of an algorithm
to determine molecular crystal structures using the Shake-and-Bake methodology.
Finally, the grid-enabled data mining approach that we introduce is able to exploit
computational cycles, which would otherwise go unused.
24.1 INTRODUCTION
The grid is a rapidly emerging and expanding technology that allows geographi-
cally distributed and independently operated resources [central processing unit (CPU)
cycles, data storage, sensors, visualization devices, and a wide variety of Internet-
ready instruments] to be linked together in a transparent fashion [1-3]. The power of
the grid lies not only in the aggregate computing power, data storage, and network
bandwidth that can be readily brought to bear on a particular problem but also on its
ease of use.