GRID COMPETITION Bioinformatics Grid Services Based On OGSA

1
GRID COMPETITION - Bioinformatics Grid Services
Based On OGSA

• Supervisors
• Mr Tan Nam Beng
• Mr Teddy Sulimin
• Members
• Ng Wee Tat
• Leong Wai Hon
• Tan Wei Siang

2
Presentation Outline

• Introduction
• Objectives Features
• Software/Technologies Used
• Grid Services Implementation
• Conclusion

3
Introduction

• OGSA - define a new common and standard
  architecture for grid-based applications
• Grid Services are an extension of Web Services
  with improved characteristics services
• Globus Toolkit 3 is a reference implementation
  that allows us to program grid-based applications
• Our project showcase the convergence of 3
  technologies ( Grid Computing, Grid Services and
  Web Services)

4
Objectives

• Developing a Web-based prototype application that
  demonstrate the necessary infrastructural
  middleware using Grid Computing and OGSA (GT3)
• Able to support bioinformatics personnel in
  performing high-throughput genomic analysis
• Develop a Distributed Blast Grid Service
• Develop a Primer Unique Pattern Search Service
• Simulate a scenario to notify/inform users of
  new and modification of protein sequences from
  Swiss Prot via SMS.

5
Features

• Java Server Pages (JSP) that provide web-based
  GUI for user interfaces
• Java Bean implementation for functionalities of
  each service
• Grid Service (OGSA) for each service running at
  the container (a server which hosts Grid
  Services)
• Broker service that determine the most available
  resource to send jobs to
• SMS Module that integrate with the Protein
  Modification Notification Grid Service for
  alerting identified personnel

6
Grid Services

• Distributed BLAST
• Unique Primer Pattern Search
• Protein Notification

7
BLAST Grid Service

• Basic Local Alignment Search Tool
• Tool that compares a input sequence against
  targeted databases
• BLAST Service provide an enhanced distributed
  Grid-based version that provide performance
  improvement
• Provide 2 mode of executions
• Allow user to choose on the machine to execute
  the job
• Allow user to run on a machine which has the most
  available resources
• Implement Memory Broker
• Leverage on Globus Toolkits MDS
• Determine the machine with the most available
  memory resources (either in term of RAM or VM) in
  the Grid network

8
BLAST Architecture

• Client access service through JSP
• Blast Grid Service is hosted by Grid container at
  the server
• Determine resource to process the job and blast
  it against targeted Sequence Database

• Advantages
• Memory Broker determine the most available
  resource which will cut down the processing time

9
Primer Service

• Primer service will determine if a input DNA
  sequence is unique within one or more fasta files
  
• Service will send jobs to different available
  resources in the Grid network
• Sending of the jobs will be done concurrently via
  the use of Java Thread
• Result will be stored in mySQL database and then
  displayed for easy future reference

10
Primer System Architecture

• Client access service through JSP
• Primer Grid Service is hosted by Grid container
  at the server
• Usage of threads allow concurrent execution of
  requests
• Server stored the results into MySQL database for
  future retrieval

• Advantages
• Support load balancing as it distribute jobs to
  available Grid Resources
• e.g 3 jobs, 2 machines
• 2 jobs ? Machine A
• 1 job ? Machine B

11
Proteins Notification

• To simulate a scenario where a (Swiss-Prot)
  protein database modification done by a scientist
  could be made known to others as soon as possible
• Making use of the Notification Service provided
  in GT3
• Client can register to our grid service so as to
  be notified (automatically by using GT3
  component)
• Use the Push Notification method in OGSA
• Upon receiving modification job request, Grid
  Service will modify protein content in database
  as well as notify other people who are interested
  in protein
• SMS message sent to user
• - Which file?
• - Which line?

12
Notification System Architecture

• Client access the Web page (Notify Grid Service)
• Grid Service is started by the Grid container at
  server
• Insertion of Notify message into the SMS Server
  via the Listener.
• Send message to user via receiving from SMS
  Server

• Advantages
• Instant SMS notification on changes to protein
  databases
• Leveraging on notification feature of OGSA

13
Conclusion

• Successully develop application that demonstrate
  the Grid Service infrastructural middleware using
  OGSA/OGSI Implementation (GT3) Globus-based
  Grid infrastructure
• Benchmarking tests on the developed framework
  illustrate performance gained for computational
  and data intensive processing
• Glad to be an early adopter of OGSA and have
  explored how the 3 technologies (Grid, Web
  Services, Grid Services) can converge to provide
  a framework for high-throughput computing
  framework