Read Between the Lines: Closing Gaps of Materials and Methods to Build Workflow from the Publication

Publishing and sharing data analysis workflow using the galaxy platform has spectacularly reduced the cost of reproducing one's research, but following the description of data analysis which had been performed by other researchers to get the exact same result is still a big challenge. To evaluate the cost of data analysis workflow from the natural language description, we have performed to rebuild the workflow of CAGE sequencing data processing done by FANTOM5 team on the galaxy platform. Though the project has already published a set of papers with a lot of supplementary of methods and online protocols, it was not that straightforward to get the same result from the raw sequencing data available in the public data repository. The results processed by the rebuilt workflow are compared with the results published online by FANTOM5 team. This case study showed that some of the important information to rebuild the workflow is missing even in the well-described documents, for example, the location of the older source code, or the parameters for command execution. As the speed of biological data production increases, it will be more important to build the framework of cost-effective research reproducibility such as an automated evaluation process of published workflow. We will provide the details of our case study, and discuss how we can assure the reproducibility with the galaxy and other possible ways to perform, share, and publish the workflow as it is "executable materials and methods".