100. Civet: A Framework for Reproducible Bioinformatics Analysis
Authors: Glen Beane (Jackson Laboratory)Al Simons (Jackson Laboratory)
Abstract: Bioinformatics analysis often combines multiple publicly available command line tools with custom tools and filters into an analysis pipeline. Analysts often script together tools into an ad hoc solution for the task at hand without designing a robust and easily maintainable pipeline. At The Jackson Laboratory we had a requirement to be able to quickly develop multiple easily maintainable, robust, and reproducible pipelines to support new clinical programs. To address this, we designed and implemented a framework, called Civet, for building bioinformatics pipelines. Civet ensures consistency and reliability for pipelines developed by different analysts throughout our organization, and because it was developed with clinical pipelines in mind, we included unique validation capabilities into Civet. Civet has the capability to verify that all the binaries, shared libraries, and reference files used in the execution of the pipeline are unchanged since the pipeline was validated.
Two-page extended abstract: pdf