Computation in big data analysis raises accessibility issues including tool choice, tool installation, choice of tool parameters and combining tools in an analysis pipeline. Addressing these issues requires advanced programming experience. Nevertheless, this is only the top of the “bioinformatics” iceberg, whose hidden part, analysis reproducibility, is still underestimated by life scientists, including bioinformaticians. Outputting an analysis in a form that ensures reproduction by other scientist (including article reviewers) requires expertise in software engineering as well as in versioning. Reproducibility of analyses can be limited by missing raw data, lack of details in the processing methods and lack of software as well as hardware details. Integrative experiments, which use multiple data sources and multiple computational tools in their analyses, exacerbate the reproducibility issue. Sharing and communication of experimental results to get them challenged and improved by peers is at the core of Science. Thus, computational biology has to face a third challenge: transparency to improve productivity and promote collaboration. In this respect, computational methods are of equal or even greater importance than text and figures. Transparency has received even less attention from life scientists than accessibility and reproducibility.
I my talk, I will discuss accessibility, reproducibility and transparency issues and show how they are addressed by the Galaxy community, through the development of a computational framework as well as the promotion of best practices in "omics" analyses.
