Report on three Genomes to Life Workshops: Data Infrastructure, Modeling and Simulation, and Protein Structure Prediction

On July 22, 23, 24, 2003, three one day workshops were held in Gaithersburg, Maryland. Each was attended by about 30 computational biologists, mathematicians, and computer scientists who were experts in the respective workshop areas The first workshop discussed the data infrastructure needs for the Genomes to Life (GTL) program with the objective to identify gaps in the present GTL data infrastructure and define the GTL data infrastructure required for the success of the proposed GTL facilities. The second workshop discussed the modeling and simulation needs for the next phase of the GTL program and defined how these relate to the experimental data generated by genomics, proteomics, and metabolomics. The third workshop identified emerging technical challenges in computational protein structure prediction for DOE missions and outlining specific goals for the next phase of GTL. The workshops were attended by representatives from both OBER and OASCR. The invited experts at each of the workshops made short presentations on what they perceived as the key needs in the GTL data infrastructure, modeling and simulation, and structure prediction respectively. Each presentation was followed by a lively discussion by all the workshop attendees. The following findings and recommendations were derived from the three workshops. A seamless integration of GTL data spanning the entire range of genomics, proteomics, and metabolomics will be extremely challenging but it has to be treated as the first-class component of the GTL program to assure GTL's chances for success. High-throughput GTL facilities and ultrascale computing will make it possible to address the ultimate goal of modern biology: to achieve a fundamental, comprehensive, and systematic understanding of life. But first the GTL community needs to address the problem of the massive quantities and increased complexity of biological data produced by experiments and computations. Genome-scale collection, analysis, dissemination, and modeling of those data are the key to success of GTL. Localizing these activities within each experimental facility that generates the data will ease integration and organization. However, integration and coordination of these activities across the facilities will be extremely critical to assure high-throughput knowledge synthesis and engage the broader biology community. Ultimately, the success of the data infrastructure will be judged by how well it is accepted by and serves the biology community