Download presentation
Presentation is loading. Please wait.
Published byAbel Moody Modified over 9 years ago
1
DNA REASSEMBLY Using Javaspace Sung-Ho Maeung Laura Neureuter
2
Overview Project Description Features of Javaspace Framework - Master/Worker Model - Design Diagram - Remote Method Invocation ( RMI ) Description of Algorithm Performance Lessons Learned Improvements Questions
3
Project Description After DNA strand has been split apart by the “shotgun” approach in order to perform sequencing, it needs to be put back together into original sequence. We used Javaspace to do this “putting back together” of the individual segments.
4
Features of Javaspace Javaspaces technology is a simple mechanism for dynamic communication, coordination, and sharing of objects between Java clients and servers. In a distributed application, Javaspace acts as a virtual space between providers and requesters of objects. One Advantage: Provides synchronization Javaspace is similar to C-Linda approach, except uses Java. Workers store and access “objects” rather than “tuples”.
5
Visual Representation of Javaspace
6
Framework – Application Patterns Many different types of patterns can be represented using Javaspace: - The Replicated-Worker Pattern - The Command Pattern - The Marketplace Pattern - others. We chose to use the Master/Worker approach to solving the reassembly problem.
7
Framework – Design Diagram
8
Framework – Remote Method Invocation (RMI) 1/3 Distributed System using Javaspace Worker as Server Master as Client RMIEntry Host name Port
9
Framework – RMI 2/3 Worker –Put RMIEntry into javaspace Host name, port Master –Get the RMIEntry from javasapce –Make RMI connection –Execute a function of the worker
10
Framework – RMI 3/3
11
Algorithm Description We parse a file into random length segments with overlaps. Each worker is assigned a random segment by the master. The rest are thrown out into javaspace. Each worker then reads a segment from javaspace and checks to see if its ends match the one it is holding. If yes, it combines the segments, and uses this new one to match against. If not, it reads another. This continues until each worker has checked all segments. (exhaustive). When all work is done, master gathers results and prints
12
Performance Evaluation
13
What We Learned JavaSpace Technology RMI Technology Javaspace is fun (but slow) Javaspace might not be the right technology to do reassembly Not a huge efficiency difference between “parallel” and “distributed” with this type of problem.
14
Improvements Algorithm needs work. Methods of synchronization need to be worked on. Load balancing Fuzzy matching implementation
15
Questions?
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.