A principled approach to reproducible research: a comparative review towards scientific integrity in computational research

Independent replication has stood up sciences in good stead, allowing for a process of self-correction and weeding out spurious claims. The factors related to inability of replication are usually related to lack of resources (e.g. money, primary data sources), uniqueness of study or a lack of time or opportunity. As the cost of irreproducible research escalates, computationally intensive research fields such as cancer research show extremely poor results in actual clinical trials pointing to acceptance of poor quality research at preclinical stage. Between complete replication and reliance purely on peer reviews, reproducible research has emerged as a common minimum, allowing regeneration of all facts and figures reported in a research. In this paper, we evaluate concepts for reproducible research, such as literate statistical programming, literate computing frameworks and scientific workflow systems along with set of practices that enable researchers create reproducible research documents. We then present a maturity model wherein reproducible research must require a minimal compliance. The paper also presents details of possible set of copyright laws and open source licenses that can be mixed to provide a compelling open access publication platform protection for researchers.