When people talk about their data infrastructure, they tend to focus on the technologies: Hadoop, Scalding, Impala, and the like. However, we’ve found that just as important as the technologies themselves are the principles that guide their use. We’d like to share our experience with one such principle that we’ve found particularly useful: reproducibility. We’ll talk about our motivation for focusing on reproducibility, how we’re using Jupyter Notebooks as our core tool, and the workflow we’ve developed around Jupyter to operationalize our approach.
We know now that much health and medical research which is published in peer-reviewed journals is wrong, and consequently much is unable to be replicated.[2-4] This is due in part to poor research practice, biases in publication, and simply a pressure to publish in order to ‘survive’. Cognitive biases that unreasonably wed to our hypotheses and results are to blame. Strongly embedded in our culture of health and medical research is the natural selection of poor science practice driven by the dependence for survival on high rates of publication in academic life. It is a classic form of cultural evolution along Darwinian lines.[6, 7] Do not think that even publications in the most illustrious medical journal are immune from these problems: the COMPare project reveals that more than 85% of large randomised controlled trials deviate seriously from their plan when the trial was registered prior to its start. An average of more than five new outcome measures was secretly added to the publication and a similar number of nominated outcomes were silently omitted. It is hardly far-fetched to propose that this drive to publish is contributing to the growth in the number of papers retracted from the literature for dubious conduct along with the increasing number of cases of research misconduct.
The biomedical research sciences are currently facing a challenge highlighted in several recent publications: concerns about the rigor and reproducibility of studies published in the scientific literature.Research progress is strongly dependent on published work. Basic science researchers build on their own prior work and the published findings of other researchers. This work becomes the foundation for preclinical and clinical research aimed at developing innovative new diagnostic tools and disease therapies. At each of the stages of research, scientific rigor and reproducibility are critical, and the financial and ethical stakes rise as drug development research moves through these stages.
Reproducibility is a foundational principle in scientific research. Yet in computational hydrology, the code and data that actually produces published results is not regularly made available, inhibiting the ability of the community to reproduce and verify previous findings. In order to overcome this problem we recommend that re-useable code and formal workflows, which unambiguously reproduce published scientific results, are made available for the community alongside data, so that we can verify previous findings, and build directly from previous work. In cases where reproducing large-scale hydrologic studies is computationally very expensive and time-consuming, new processes are required to ensure scientific rigour. Such changes will strongly improve the transparency of hydrological research, and thus provide a more credible foundation for scientific advancement and policy support.
In a recent Opinion article, Parker et al.  highlight a range of important issues and provide tangible solutions to improve transparency in ecology and evolution (E&E). We agree wholeheartedly with their points and encourage the E&E community to heed their advice. However, a key issue remains conspicuously unaddressed: Parker et al. assume that ‘deliberate dishonesty’ is rare in E&E, yet evidence suggests that occurrences of scientific misconduct (i.e., data fabrication, falsification, and/or plagiarism) are disturbingly common in the life sciences .
The scientific community is increasingly concerned with cases of published "discoveries" that are not replicated in further studies. The field of mouse phenotyping was one of the first to raise this concern, and to relate it to other complicated methodological issues: the complex interaction between genotype and environment; the definitions of behavioral constructs; and the use of the mouse as a model animal for human health and disease mechanisms. In January 2015, researchers from various disciplines including genetics, behavior genetics, neuroscience, ethology, statistics and bioinformatics gathered in Tel Aviv University to discuss these issues. The general consent presented here was that the issue is prevalent and of concern, and should be addressed at the statistical, methodological and policy levels, but is not so severe as to call into question the validity and the usefulness of the field as a whole. Well-organized community efforts, coupled with improved data and metadata sharing were agreed by all to have a key role to play in view of identifying specific problems, as well as promoting effective solutions. As replicability is related to validity and may also affect generalizability and translation of findings, the implications of the present discussion reach far beyond the issue of replicability of mouse phenotypes but may be highly relevant throughout biomedical research.