An analysis of tool requirements for collaborative annotation

In the last years, the amount of genomic data produced has increased significantly. These data need to be processed and linked to biological information in order to make a meaningful use of them. Automated tools support these tasks, but their results have to be curated by experts, what creates a bottleneck. Although some collaborative community annotation efforts have attempted to address this issue, they have not been overly successful. This paper carries out a social analysis of them and their tools through the Activity Theory framework, looking for limitations and potential solutions. Its conclusions are used to propose requirements for an actual annotation tool, MASSA (Multi-Agent System to Support functional Annotation), which works in junctional annotation (i.e., assigning biological functions). The case study reports a preliminary evaluation of their implementation and the analysis of the tool performance according to a state-of-the-art benchmark.