Research data management for bioimaging: the 2021 NFDI4BIOIMAGE community survey

Abstract

Background: Knowing the needs of the bioimaging community with respect to research data management (RDM) is essential for identifying measures that enable adoption of the FAIR (findable, accessible, interoperable, reusable) principles for microscopy and bioimage analysis data across disciplines. As an initiative within Germany's National Research Data Infrastructure, we conducted this community survey in summer 2021 to assess the state of the art of bioimaging RDM and the community needs. Methods: An online survey was conducted with a mixed question-type design. We created a questionnaire tailored to relevant topics of the bioimaging community, including specific questions on bioimaging methods and bioimage analysis, as well as more general questions on RDM principles and tools. 203 survey entries were included in the analysis covering the perspectives from various life and biomedical science disciplines and from participants at different career levels. Results: The results highlight the importance and value of bioimaging RDM and data sharing. However, the practical implementation of FAIR practices is impeded by technical hurdles, lack of knowledge, and insecurity about the legal aspects of data sharing. The survey participants request metadata guidelines and annotation tools and endorse the usage of image data management platforms. At present, OMERO (Open Microscopy Environment Remote Objects) is the best known and most widely used platform. Most respondents rely on image processing and analysis, which they regard as the most time-consuming step of the bioimage data workflow. While knowledge about and implementation of electronic lab notebooks and data management plans is limited, respondents acknowledge their potential value for data handling and publication. Conclusion: The bioimaging community acknowledges and endorses the value of RDM and data sharing. Still, there is a need for information, guidance, and standardization to foster the adoption of FAIR data handling. This survey may help inspiring targeted measures to close this gap.

Keywords: FAIR-principles; OMERO; Research data management; bioimage analysis; bioimaging; microscopy.

Figure 1.. Overview of the survey respondents.

Shown are responses for (a) “I work at/I am affiliated with”, (b) “My current (primary) position is located in”, and (c) “My current primary position is”. The latter criterion was used to distinguish four different groups i) undergraduate and PhD students, ii) postdoctoral and permanent-term researchers, iii) junior and senor group leaders, and iv) research support and facility staff. n = 198, five respondents are not included (two stated “Consultant” and “Company”, three left the field blank). The symbols used in c represent these groups throughout the manuscript. d) Participants were asked to state which approaches describe their work best (multiple answers possible). e) Number of participants stating to use, not to use, or to be unsure if they use bioimaging methods in their work.

Figure 2.. Knowledge and use of bioimaging methods.

a) Participants were asked to state, if their work includes the indicated methods (± 12 months) with a preselected list and a free-text option. This question was shown to respondents who stated “I use bioimaging or biophotonics methods” or "I am not sure, if one of my methods is a bioimaging or biophotonics method”, incl. one person who left the field blank ( Figure 1e, n = 188). b) Participants could choose one method as the most important for their work (Suppl. Figure 3 ), for which we asked in which aspect(s) of the method they are involved (blue bars, multiple choice) and which step is the most time-consuming (red bars, single choice). c) Participants were asked about their main information source(s) for learning a new bioimaging method (multiple choice). See also Suppl. Figure 4. Abbreviations: AFM Atomic Force Microscopy; CLEM Correlated Light and Electron Microscopy; FLIM Fluorescence Lifetime Imaging; FRAP Fluorescence Recovery After Photobleaching; FRET Förster Resonance Energy Transfer; SPIM Selective Plane Illumination Microscopy; REM Reflection Electron Microscopy; TEM Transmission Electron Microscopy; TIRF Total Internal Reflection.

Figure 3.. Role of bioimage analysis and aspects of autonomous processing and analysis.

a) Mode of image analysis by autonomous users and (b) their sources information for learning bioimage analysis procedures (n = 168). c) Self-reported skill levels in bioimage analysis of autonomous users, and (d, e) comparison between skilled professionals/qualified experts (n = 42) vs. inexperienced and beginners (n = 40).

Figure 4.. Data management platform knowledge and use by participants.

We presented a list of generic and image-data-tailored management systems and asked respondents for their use, interest, and knowledge about each system on the indicated scale. Abbreviations: CATMAID Collaborative Annotation Toolkit for Massive Amounts of Image Data; FAIRDOM-SEEK Findable, Accessible, Interoperable and Reusable Data, Operating procedures and Models; iRODS Integrated Rule-Oriented Data System; OMERO Open Microscopy Environment Remote Objects; OpenBIS Open Biology Information System.

Figure 5.. The role of metadata for research data management and the needs for metadata annotation.

a) Respondents stated their opinion about three statements on metadata (“The meaning of the term “metadata” is clear to me”, “Systematic and exhaustive metadata annotation is essential for data management in a research project”, and “Systematic and exhaustive metadata annotation is easy and timesaving”) on a five-item scale. The bar graphs show the relative fraction per answer-item in each of the four groups (undergraduate and PhD students, n = 55; Postdoctoral and permanent researchers, n = 66; junior and senior group leaders/professors, n = 27; Research support staff, n = 48). b) Participants were asked to state up to three most urgent needs to improve metadata handling and annotation. Abbreviation: ELN Electronic Lab Notebook.

Figure 6.. Knowledge and use of Data Management Plans (DMPs) and Electronic Lab Notebooks (ELNs).

a) Answers of participants from the four career level groups (undergraduate and PhD students, n = 55; postdoctoral and permanent researchers, n = 66; junior and senior group leaders/professors, n = 27; research support staff, n = 50) to the statement “I know what a DMP is and what it is used for”. The answers to the right-hand statements (grey background) are only shown for participants who stated to know what a DMP is (84 respondents from career level groups, 2 respondents who stated “Company” and “Consultant” as their career level). b) Answers of all participants to the indicated statements about ELNs.

Figure 7.. RDM knowledge and the FAIR (finable, accessible, interoperable, reusable) principles.

a, b) Respondents provided their opinion about presented statements on a five-item scale. See Suppl. Figure 11 for the full set of questions. Shown are the relative fraction per answer item in each group (undergraduate and PhD student, n = 55; Postdoctoral and permanent researchers, n = 66; Junior and senior group leaders, n = 27; Research support and facility staff, n = 50). c) Respondents were asked about their adoption and knowledge of the FAIR principles on the indicated five-item answer scale.

Figure 8.. Experience with and opinion about bioimage data sharing.

a, b, c) Shortened statements about private or public data sharing and data reuse. Shown are the relative answer distributions of the five agreement levels for each of the analyzed groups (undergraduate and PhD students, n = 55; Postdoctoral and permanent researchers, n = 66; junior and senior group leaders/professors, n = 27; Research support staff, n = 48). d) Opinions on three statements about repositories. e) Knowledge and use of preselected data repositories including bioimaging-specific, research-area-specific, and generic repositories. See Suppl. Figure 12 for the full set of questions corresponding to this figure. Abbreviations: BIA BioImage Archive; EMPIAR Electron Microscopy Public Image Archive; IDR Image Data Resource; PGP Plant Genomic and Phenomics; YRC PIR Yeast Research Center Public Image Repository.

Figure 9.. Proper handling of large-scale, complex data as frequently acquired in bioimaging is a challenge for researchers, data providers, and data users.

Targeted measures must rely on a firm knowledge of the community perspective and its needs to transform the research data management whirlwind into a well-managed bioimage data life cycle (cartoon produced by Henning Falk for this article and published with permission).