Abstract
Summary: We introduce Tibanna, an open-source software tool for automated execution of genomic pipelines on Amazon Web Services (AWS). Tibanna accepts portable pipeline standards including Common Workflow Language (CWL) and Docker, and it can be used through a simple command line interface (CLI) or Python Application Programming Interface (API). It adopts a strategy of isolation and optimization of individual executions, without pre-configured clusters. Pipelines are automatically downloaded, executed and monitored on a single local command. The versatility of Tibanna makes it ideal both for Data Commons, where data to be processed come in sparsely or in bursts, and for stand-alone large-scale parallel batch processing. Tibanna has been used to reproducibly process terabytes of genomic data for the 4D Nucleome (4DN) Network. Availability: Source code is available on GitHub at https://github.com/4dn-dcic/tibanna. Contact: peter_park@hms.harvard.edu
