Reproducibility is a fundamental requirement of the scientific process since it enables outcomes to be replicated and verified. Computational scientific experiments can benefit from improved reproducibility for many reasons, including validation of results and reuse by other scientists. However, designing reproducible experiments remains a challenge and hence the need for developing methodologies and tools that can support this process. Here, we propose a conceptual model for reproducibility to specify its main attributes and properties, along with a framework that allows for computational experiments to be findable, accessible, interoperable, and reusable. We present a case study in ecological niche modeling to demonstrate and evaluate the implementation of this framework.
Geoinformatics deals with spatial and temporal information and its analysis. Research in this field often follows established practices of first developing computational solutions for specific spatiotemporal problems and then publishing the results and insights in a (static) paper, e.g. as a PDF. Not every detail can be included in such a paper, and particularly, the complete set of computational steps are frequently left out. While this approach conveys key knowledge to other researchers it makes it difficult to effectively re-use and reproduce the reported results. In this vision paper, we propose an alternative approach to carry out and report research in Geoinformatics. It is based on (computational) reproducibility, promises to make re-use and reproduction more effective, and creates new opportunities for further research. We report on experiences with executable research compendia (ERCs) as alternatives to classic publications in Geoinformatics, and we discuss how ERCs combined with a supporting research infrastructure can transform how we do research in Geoinformatics. We point out which challenges this idea entails and what new research opportunities emerge, in particular for the COSIT community.
This data article introduces a reproducibility dataset with the aim of allowing the exact replication of all experiments, results and data tables introduced in our companion paper (Lastra-Díaz et al., 2019), which introduces the largest experimental survey on ontology-based semantic similarity methods and Word Embeddings (WE) for word similarity reported in the literature. The implementation of all our experiments, as well as the gathering of all raw data derived from them, was based on the software implementation and evaluation of all methods in HESML library (Lastra-Díaz et al., 2017), and their subsequent recording with Reprozip (Chirigati et al., 2016). Raw data is made up by a collection of data files gathering the raw word-similarity values returned by each method for each word pair evaluated in any benchmark. Raw data files was processed by running a R-language script with the aim of computing all evaluation metrics reported in (Lastra-Díaz et al., 2019), such as Pearson and Spearman correlation, harmonic score and statistical significance p-values, as well as to generate automatically all data tables shown in our companion paper. Our dataset provides all input data files, resources and complementary software tools to reproduce from scratch all our experimental data, statistical analysis and reported data. Finally, our reproducibility dataset provides a self-contained experimentation platform which allows to run new word similarity benchmarks by setting up new experiments including other unconsidered methods or word similarity benchmarks.
Background:The complex nature of biological data has driven the development of specialized software tools. Scientificworkflow management systems simplify the assembly of such tools into pipelines, assist with job automation, and aidreproducibility of analyses. Many contemporary workflow tools are specialized or not designed for highly complexworkflows, such as with nested loops, dynamic scheduling, and parametrization, which is common in, e.g., machinelearning.Findings:SciPipe is a workflow programming library implemented in the programming language Go, for managingcomplex and dynamic pipelines in bioinformatics, cheminformatics, and other fields. SciPipe helps in particular withworkflow constructs common in machine learning, such as extensive branching, parameter sweeps, and dynamicscheduling and parametrization of downstream tasks. SciPipe builds on flow-based programming principles to supportagile development of workflows based on a library of self-contained, reusable components. It supports running subsets ofworkflows for improved iterative development and provides a data-centric audit logging feature that saves a full audit tracefor every output file of a workflow, which can be converted to other formats such as HTML, TeX, and PDF on demand. Theutility of SciPipe is demonstrated with a machine learning pipeline, a genomics, and a transcriptomics pipeline.Conclusions:SciPipe provides a solution for agile development of complex and dynamic pipelines, especially in machinelearning, through a flexible application programming interface suitable for scientists used to programming or scripting.
Shared scientific resources, also known as core facilities, support a significant portion of the research conducted at biomolecular research institutions. The Association of Biomolecular Resource Facilities (ABRF) established the Committee on Core Rigor and Reproducibility (CCoRRe) to further its mission of integrating advanced technologies, education, and communication in the operations of shared scientific resources in support of reproducible research. In order to first assess the needs of the scientific shared resource community, the CCoRRe solicited feedback from ABRF members via a survey. The purpose of the survey was to gain information on how U.S. National Institutes of Health (NIH) initiatives on advancing scientific rigor and reproducibility influenced current services and new technology development. In addition, the survey aimed to identify the challenges and opportunities related to implementation of new reporting requirements and to identify new practices and resources needed to ensure rigorous research. The results revealed a surprising unfamiliarity with the NIH guidelines. Many of the perceived challenges to the effective implementation of best practices (i.e., those designed to ensure rigor and reproducibility) were similarly noted as a challenge to effective provision of support services in a core setting. Further, most cores routinely use best practices and offer services that support rigor and reproducibility. These services include access to well-maintained instrumentation and training on experimental design and data analysis as well as data management. Feedback from this survey will enable the ABRF to build better educational resources and share critical best-practice guidelines. These resources will become important tools to the core community and the researchers they serve to impact rigor and transparency across the range of science and technology.
Scientific data reuse requires careful curation and annotation of the data. Late stage curation activities foster FAIR principles which include metadata standards for making data findable, accessible, interoperable and reusable. However, in scientific domains such as biomolecular nuclear magnetic resonance spectroscopy, there is a considerable time lag (usually more than a year) between data creation and data deposition. It is simply not feasible to backfill the required metadata so long after the data has been created (anything not carefully recorded is forgotten) – curation activities must begin closer to (if not at the point of) data creation. The need for foregrounding data curation activities is well known. However, scientific disciplines which rely on complex experimental design, sophisticated instrumentation, and intricate processing workflows, require extra care. The knowledge gap investigated by this research proposal is to identify classes of important metadata which are hidden within the tacit knowledge of a scientist when constructing an experiment, hidden within the operational specifications of the scientific instrumentation, and hidden within the design / execution of processing workflows. Once these classes of hidden knowledge have been identified, it will be possible to explore mechanisms for preventing the loss of key metadata, either through automated conversion from existing metadata or through curation activities at the time of data creation. The first step of the research plan is to survey artifacts of scientific data creation. That is, (i) existing data files with accompanying metadata, (ii) workflows and scripts for data processing, and (iii) documentation for software and scientific instrumentation. The second step is to group, categorize, and classify the types of "hidden" knowledge discovered. For example, one class of hidden knowledge already uncovered is the implicit recording of data as its reciprocal rather than the value itself, as in magnetogyric versus gyromagnetic ratios. The third step is to design/propose classes of solutions for these classes of problems. For instance, reciprocals are often helped by being explicit with units of measurement. Careful design of metadata display and curation widgets can help expose and document tacit knowledge which would otherwise be lost.