Finding Protein Binding Sites Using Volunteer Computing Grids

Abstract

This paper describes initial work in the development of the DNA@Home volunteer computing project, which aims to use Gibbs sampling for the identification and location of DNA control signals on full genome scale data sets. Most current research involving sequence analysis for these control signals involve significantly smaller data sets, however volunteer computing can provide the necessary computational power to make full genome analysis feasible. A fault tolerant and asynchronous implementation of Gibbs sampling using the Berkeley Open Infrastructure for Network Computing (BOINC) is presented, which is currently being used to analyze the intergenic regions of the Mycobacterium tuberculosis genome. In only three months of limited operation, the project has had over 1,800 volunteered computing hosts participate and obtains a number of samples required for analysis over 400 times faster than an average computing host for the Mycobacterium tuberculosis dataset. We feel that the preliminary results for this project provide a strong argument for the feasibility and public interest of a volunteer computing project for this type of bioinformatics.