Hardening Globus Genomics to make large-scale genomics analysis available for everybody

点击翻译按钮获取中文摘要

Project Summary Globus Genomics has been developed at the Computation Institute, University of Chicago as an advanced genomics analysis platform running as a Software-as-a-Service on Amazon Web Services, powered by Globus and Galaxy. It was developed to meet the needs of both researchers and core lab providers who require a high-quality service with state-of-the-art capabilities to help streamline data movement, simplify the creation of genomics analysis pipelines, automate the execution of those pipelines, and run analysis at very large scale on elastic compute infrastructure. Globus Genomics, under development for three years, has been used extensively by researchers at leading institutions, including University of Washington, University of Chicago, Washington University St. Louis, Georgetown University, and Johns Hopkins. There is significant potential and demand to expand use of the service to meet the rapidly growing genomics analysis needs of both existing users and large communities of new users. We now propose work that will amplify the utility and impact of Globus Genomics by providing (1) scalability to 1000s of simultaneous analyses by 1000s of users, (2) support for state-of-art high performance workflows and tools, including large-scale imputation analysis and consensus calling on structural variants, (3) automated cost and performance optimization to slash cloud computing costs and turnaround times, and (4) powerful dashboards for end-to-end and summary views of large-scale analyses. These enhancements will be enabled by development in the following key areas: enhancing and extending the Globus Genomics computational framework to enable high-performance reliable execution of standard and novel NGS analysis workflows on large and extremely large datasets; creating and maintaining state-of-the-art pipelines for variant calling, whole genome analysis, RNASeq and ChipSeq, which involves computationally profiling the latest versions of tools and understanding different computational modalities for optimal execution on Amazon Web Services; creating a profiling and optimization framework to enable automated, cost- and/or time-optimal configuration of NGS applications and workflows on large cloud systems; and creating an automatic computational provisioning framework. The grant award would allow us to address key needs of current and prospective users and thus to provide an important bioinformatics platform to researchers who otherwise could not easily access such capabilities.