D istrib u ted G enom e

Genome projects are very demanding undertakings and are often carried out collaboratively by multiple research centers There are many di erent types of tasks that must be performed both by re searchers and automated tools These include such activities as shot gun sequencing sequence nishing sequence similarity searches data annotation oligonucleotide synthesis library construction and data submission These individual tasks are organized into work ows that carry out a particular function such as providing an annotated se quence of a cosmid clone The individual tasks may be carried out at one or more of the participating institutions A single work ow may be spread across multiple research centers Creating software systems to support distributed work ows presents developers with a number of challenges such as coordinating the execution of applications run ning on di erent systems transporting data between systems inte grating legacy applications providing recovery mechanisms and cre ating user interfaces Additionally there will likely be frequent change to the organizational procedures especially in the early stages of a genome project This paper discusses using a general purpose work ow management system WfMS to address these challenges in the implementation of an information system that manages a geographi cally distributed genome project A prototype application built with the METEOR WfMS is described which is running on several systems at the University of Georgia