PRESENTERS: Hilmar Lapp, Karen Cranston, Mine Çetinkaya-Rundel, Mark Delong, Erich Huang, Dan Leehr, Darin London, Paul Magwene
DEPARTMENTS: Center for Genomic and Computational Biology (GCB)
FORMAT: Panel Discussion (Moderator and panel members discussing a topic andor taking questions from the audience.)
Reproducibility is a cornerstone of science, enabling scientists to “stand on the shoulders of giants.” Computational research, which is increasingly pervasive as ever more of science becomes data rich, faces particular challenges to reproducibility, such as dependency chains, runtime environment heterogeneity, difficulty of end-to-end automation and resource demands. Many of these challenges are technical, and overcoming them therefore inherently involves technologies, tools and practices that often have little to do with domain science. Nonetheless, computationally savvy domain scientists undeterred by the bewildering technology soup have pushed the envelope in improving the reproducibility of computational research. Yet, in part because of the technology soup, there are huge opportunities for campus IT groups and organizations to take an active, leading role in the intelligent and coordinated development, improvement, deployment and teaching of these technologies and practices. This panel brings together faculty, bioinformaticians and IT personnel from across Duke who are participating in an international collaboration to develop a reproducible science curriculum, and who are actively prototyping and developing tools to improve the reproducibility of data and compute-intensive research. In summary, we hope to position Duke to become one of the leaders and innovators of reproducible science by stimulating dialog and collaboration between the research, computational and IT talents on campus.