Handling Large Scale Biological Workloads on Networked Computing Systems 
   - Potential for Divisible Load Paradigm

Bharadwaj Veeravalli
National University of Singapore, Singapore

Abstract:

In this talk, I will first give a quick overview of Divisible Load
Theory (DLT) - the model, some key results, and highlight on
applications that had directly used this paradigm. This will be the
first part of the talk. In the second part of the talk, I will present
few of our recent works that had directly used DLT paradigm for
biological sequence matching problems.

Similarity matching or homology detection is an impetrative step in
many of the Bioinformatics applications.  For instance, protein
structure-to-structure comparison methods invariably start with some
understanding on the results from similarity detection
mechanisms. Protein classification problems based on domain specific
knowledge invariably use sequence aligning methods in a fairly
repetitive way thus consuming enormous amount of computing power from
resources. These are few examples to immediately concede the
computational challenge involved in handling such voluminous
data. Modern day computing platforms, such as networked workstations,
a high-speed LAN infrastructure or even Grid computing domain offers
enormous amount of power that can be tapped for handling such data
intensive applications. Such applications seem to demand
'divide-and-conquer' approach to gain significant speed-ups, although
the data partitioning and distribution among the nodes play a vital
role in deciding the speed-up. I will present our experiences and
outcomes in adopting DLT paradigm for biological problems on two
different topologies - Bus and Mesh networks. The latter topology is
attempted to design an automatic computational engine for handling
fairly large number of sequences delivering an impressive speed-up!
These ventures clearly encourage such applications to consider
modeling under divisible load paradigm and also serve as a viable
starting point towards mapping onto high-performance systems.


Brief Biography:

Bharadwaj Veeravalli, Member, IEEE & IEEE-CS, received his Bachelors
Degree in Science(Physics), from Madurai-Kamaraj Uiversity, India in
1987, Master's in Electrical Communication Engineering from Indian
Institute of Science, Bangalore, India in 1991 and PhD from Department
of Aerospace Engineering, Indian Institute of Science, Bangalore,
India in 1994. He did his post-doctoral research in the Department of
Computer Science, Concordia University, Montreal, Canada, in 1996. He
is currently with the Department of Electrical and Computer
Engineering, Communications and Information Engineering (CIE)
division, at The National University of Singapore, Singapore, as a
tenured Associate Professor. His main stream research interests
include, Cluster/Grid computing, Scheduling in parallel and
distributed systems, Bioinformatics & Computational Biology,
Multiprocessor systems , and Multimedia computing. He is one of the
earliest researchers in the field of divisible load theory (DLT). He
has published over 90 papers in high-quality International Journals
and Conferences. He has co-authored three research monographs in the
areas of PDS, Distributed Databases(competitive algorithms), and
Networked Multimedia Systems, in the years 1996, 2003, and 2005,
respectively. He had guest edited a special issue on Cluster/Grid
Computing for IJCA, USA journal in 2004. He is currently serving the
Editorial Board of IEEE Transactions on Computers, IEEE Transactions
on SMC-A, and International Journal of Computers & Applications, USA,
as an Associate Editor.  Bharadwaj Veeravalli's academic profile,
professional activities, main stream and peripheral research
interests, research projects and collaborations, most recent list of
publications, can be found via: http://cnds.ece.edu.sg/elebv