Biocuration - mapping resources and needs

Introduction

Together with bioinformatics, curated databases have become an essential part of modern molecular biology. Databases such as Ensembl¹, COSMIC² and PomBase³ are accessed daily by thousands of people worldwide, illustrating that well-structured life science data, available to all in public repositories, are fundamental to research. Often, these biological databases are annotated by biocurators whose work, in a simplified explanation, is to i) collect scientific data, ii) verify and validate the information collected, iii) add value by structuring it in a logical, consistent and relevant manner and iv) integrate it into databases. Bourne and McEntyre⁴, in their homage to the profession, considered biocurators as the “museum cataloguers of the Internet age”. We believe biocurators are more than this, they are integral to the modern life sciences and pivotal to good data management. They are the guardians of the integrity and FAIRness of life sciences data.

The art and science of biocuration helps make data Findable, Accessible, Interoperable and Reusable (FAIR)⁵; it places data into context, makes it more interoperable, adding unique identifiers, licences, and structured descriptions alongside other appropriate metadata. In this respect, biocuration shares some aspects of data stewardship, but there are also some differences. Data stewardship involves improving the FAIRness of a dataset, but doesn’t include adding value to the data, via curation and integration. In many ways, biocuration begins where data stewardship finishes, with data stewards helping researchers to prepare their data for biocurators to add value through integration into appropriate databases.

The European life sciences infrastructure for biological information (ELIXIR; https://elixir-europe.org) is an inter-governmental organisation that brings together bioinformatics resources across Europe and helps researchers to find, analyse, and exchange life science data. In 2016, the ELIXIR Data Platform put in place a process to identify European data resources that are of fundamental importance to research in the life sciences⁶. Many of these resources are manually curated and ELIXIR recognises the substantial added value that this brings to these resources. Manual biocuration is a highly valued task, and the biocuration efforts of a resource are an important indicator of its quality.

The International Society for Biocuration (ISB; https://www.biocuration.org) counts over 250 members working in over 82 institutions around the globe. The 27th annual Nucleic Acids Research database issue and molecular biology database collection lists 1613 databases, including 59 new databases⁷. Biocurators not only work in well-known databases but also in ad-hoc, often behind the scenes projects where biocuration is needed, be they in academia, non-governmental organisations or private companies. A large, diverse body of individuals and teams are involved in biocuration across ELIXIR. It is anticipated that a number of these may not identify themselves as biocurators per se – this could be due to them being unaware of this as a position or career path, or that biocuration is not the main aspect of their job. In this article, we use a broad definition of ‘biocurator’ to encompass anyone who carries out biocuration tasks as part of their work regardless of whether it is the primary focus of their role.

The ELIXIR Training Platform (TrP), one of the five infrastructure pillars of ELIXIR, aims to strengthen national bioinformatics programmes, grow bioinformatics capacity and competences across ELIXIR member states and empower researchers to use ELIXIR’s services and tools (https://elixir-europe.org/platforms/training). A large number of biocurators work within the ELIXIR Nodes, and the ELIXIR TrP requires a clear picture of who they are, their profiles, what resources they work on, the tools that they use and, in particular, their training needs. The last survey of biocurators was conducted several years ago in 2011, and in an international context where ELIXIR and the Global Organization for Bioinformatics Learning, Education and Training (GOBLET, https://www.mygoblet.org) did not yet exist. An up-to-date profile of the landscape of biocuration would allow these organisations and employers to take action to recognise and value biocuration, and to plan future training.

In this research article we describe the outcomes of an ELIXIR Biocuration Implementation Study that aims to i) identify communities of biocurators within ELIXIR and worldwide, ii) review and map the kind of biocuration work being done, for which databases and life science/health domains, iii) assess the capacity requirements for new biocurators and training provision available, and iv) draw a picture of the biocuration career development. We outline a set of recommendations to ensure that the biocuration career path can be even more valued. We urge the community to ensure that the knowledge and skills offered by existing biocurators are shared across ELIXIR Nodes and beyond, and to develop training so it is available to those areas and disciplines where biocurators needs are in highest demand.

Methods

Results

Communities of biocurators within ELIXIR and worldwide

The global survey received 212 responses with most from Europe and the USA (Figure 1).

Figure 1. Countries in which biocurators work.

The map represents the countries in which those biocurators who responded to the survey work. The colour shading in the map indicates the number of respondents from each country.

Respondents work in 91 different organisations, with 84% of respondents belonging to academic institutions. Eight percent of the respondents work in an industrial organisation and 1% span both academia and industry (see Underlying data: Global_survey_deidentified⁸). Among the named academic institutes, 25% of respondents belong to international biocuration hubs, including EMBL-EBI, the Wellcome Sanger Institute and SIB. In free text answers, respondents indicated that they work across a wide range of life science domains, with the most commonly mentioned being bioinformatics, genomics, genetics, biology and biochemistry (Figure 2).

Figure 2. The variety of life science domains that biocurators belong to.

The size of the text in the word cloud is related to how often particular life science domains were mentioned by respondents. The colours were chosen to create a better contrast between the words.

The survey also shows that the job titles of respondents are diverse, including uncommon titles, such as knowledge engineer, data editor, repository manager, and data wrangler (the diversity of job titles is reflected in the word cloud Figure 3A). Even for biocuration hubs, such as the EMBL-EBI, a large variety of job titles was reported (Figure 3B).

Figure 3. Job titles provided by respondents of the global survey.

The figure shows all responses (A) and responses filtered by respondents from EMBL-EBI (B). The size of the text in the word cloud is related to how often particular job titles were mentioned by respondents. The colours were chosen to create a better contrast between the words.

This diversity of job titles makes biocurator positions difficult to identify, and this may impede the building of a sustainable community. In both the guided conversations and workshops, biocurators emphasised that the inconsistency of job titles hampers their career planning. Biocurators may not easily find relevant job adverts or in fact identify with them based on the title alone, and employers may have difficulty in determining what skills and knowledge to associate with which job title.

Biocuration training

In our conversations with biocurators and at the workshops, biocurators emphasised that it can take a long time to be trained as a biocurator and most agreed that biocuration is a continuous learning process. According to the survey, 46 of 164 (28%) respondents said that they have received training in biocuration, while the rest said that they have not received any training (see Underlying data: Global_survey_deidentified⁸). Training was mainly received in the form of one-to-one teaching and self-directed training, or via face-to-face courses (Figure 4). Respondents view one-to-one and self-directed training as the most impactful forms. The in-house nature of this training combined with the specific nature of biocuration tasks suggests that biocuration knowledge gained in one role is not necessarily easily transferable to another.

Figure 4. The type of training survey respondents had access to.

Respondents could choose several answers. This question was answered by 44 respondents. The percentage of respondents who chose the respective training type is also indicated in the figure.

From the responses provided, we collated a list of face-to-face and online training courses, materials and providers (see Underlying data: Biocuration training course list⁸), which may be relevant to biocurators. This course collection is not exhaustive and may again reflect the nature of the informal 1:1 and in-house training many respondents said they had received.

In the survey, from free-text answers, the most commonly indicated training needs are programming/scripting/coding (mentioned in 18 of 134 responses - 13%), development and use of ontologies (mentioned in 16 of 134 responses - 12%), and database management (mentioned in 10 of 134 responses - 7%) (see Underlying data: Global_survey_deidentified⁸). Survey respondents and biocurators that we spoke to highlighted that computational skills, such as programming/scripting/coding and extracting relevant data from the literature, are decisive for a successful career development yet are the most difficult to learn. Other skills that were highlighted as being essential for a successful biocuration career include biological knowledge (mentioned in 41 of 124 responses - 33%), attention to detail (mentioned in 23 of 124 responses - 19%), patience (mentioned in 11 of 124 - 9%), communication (mentioned in 8 of 124 responses - 6%) and curiosity (mentioned in 8 of 124 responses - 6%) (see Underlying data: Global_survey_deidentified⁸). These views were reiterated and emphasised by biocurators that we spoke to individually and at the workshops.

Discussion and recommendations

Modern science, whatever the discipline, is turning into something of a data science. This seems particularly true of the life sciences, where the explosion in omics data, followed by an increase in publications and linked datasets, has led to a massive overall increase in freely available structured and unstructured biological data. Biocuration, the action of combining data from multiple sources, and adding value by converting them into a coherent, consistent and structured interoperable dataset, has never been more essential to enable the understanding, digestion and interpretation of these data.

In this ELIXIR Implementation Study, we asked biocurators from across the world for their thoughts on the state of biocuration, both in terms of the day-to-day technicalities - what tools they use, what skills they need - and the more long range subjects of training and career opportunities. We acknowledge that the survey may not have reached all biocuration communities due to, for example, the preference of certain social media channels in different countries or the lack of awareness of ISB and ELIXIR in certain (geographical) communities. Nevertheless, we believe that the results are a fair representation of the varied views and experiences of biocurators worldwide. Our study highlights the diversity of roles and experience within the biocuration community but also reinforces the need for improved recognition of this work and career.

The day-to-day activities of biocurators are amazingly diverse and include both biocuration and non-biocuration tasks such as training, outreach, website design and project management. As noted by Samili and Vita¹³, this highlights that biocurators possess many transferable skills in addition to their scientific and biocuration expertise, characteristics that are highly valued by employers. Our survey shows that biocurators in different resources work quite differently, using different tools and following different workflows and pipelines. There are some efforts to encourage biocurators in different resources to streamline their work by following the same biocuration workflows¹⁴. Steps have been taken to encourage this streamlining through the addition of biocuration-related databases, curation tools and standard metadata records to FAIRsharing^9,11 (an ELIXIR Recommended Interoperability Resource and the ELIXIR Registry of Standards). However, it is clear from our survey that the diversity of biocuration practices and the data curated mean that a universal biocuration tool or workflow are still some time away.

Having said that, some commonalities in the experience of biocurators can be found. For instance, the clear need for more training on programmatic or coding skills, such as Python, and data modelling skills such as the creation, maintenance and use of ontologies. This is in line with a 2011 survey of biocurators carried out by the ISB in which 55% of respondents thought that better training in computer languages would be beneficial for their jobs, and 43% indicated that they would benefit from better training in bioinformatics¹⁵. Similarly, in response to a 2017 survey of biocuration training needs conducted during the development of the Postgraduate certificate in Biocuration at the University of Cambridge, 32% of respondents rated ‘Application of programming to curation tasks - e.g. scripting in Python as their top training need (S.L. Morgan, EMBL European Bioinformatics Institute; personal communication). From our discussions, it became clear to us that it is essential for biocurators to learn coding skills, not only to be able to wrangle data and perform tasks more efficiently but to facilitate discussion with database developers and gain a greater understanding of how the data they curate is represented in a database.

In our survey, 28% of respondents indicated that they had received biocuration training and some training courses were specifically named by respondents. It is possible that this relatively low figure is due to the way in which respondents interpreted the question. For example, respondents might not necessarily equate “training in a programming language”, “1:1 training” or “self-directed training” as “biocuration training”. However, a low level of formal training was also reported in response to the survey of biocuration training needs conducted during the development of the Postgraduate certificate in Biocuration at the University of Cambridge. In that survey, 12% of respondents had received training in the form of formal short courses while 88% had received informal 1:1 training in work (S.L Morgan, personal communication). The specific nature of biocuration tasks observed in our survey calls for a shift in the way that we teach biocuration skills. There is a need to develop bioinformatics training that is targeted to biocurators (e.g. programming skills for biocuration related tasks vs programming for research related tasks). Ensuring that all biocurators are up to speed with the latest biocuration and data stewardship best practices should increase the standard and consistency of biocuration while providing transferable skills that improve career mobility.

The lower reported level of formal training suggests that a potential challenge for biocurators is in being able to find available and/or appropriate training courses. An absence of training courses will also affect community building, with fewer opportunities for biocurators to meet, network and share expertise. To facilitate expanded training for biocurators, training providers should make their courses and training materials easier to find, for example by sharing them in portals such as ELIXIR’s TeSS (the ELIXIR Training and Events Portal; https://tess.elixir-europe.org/)¹², and by adopting FAIR principles to ensure that they are appropriately tagged and described¹⁶. This would allow democratisation of training efforts such that they expand out from centres of excellence (such as the EMBL-EBI or SIB) to independent biocurators. Some training events and materials listed in the survey have already been indexed in TeSS¹², which will allow the identification of gaps in training provision.

Our study also highlights the need for improved career support. This sentiment seems to have changed little from the 2011 ISB survey, in which 82% of respondents felt concerned about future work opportunities¹⁵. It can take many years of training and experience to become a biocurator, yet currently, salaries and more importantly job security, often do not reflect this investment. Key barriers to career progression highlighted in our survey are the diversity of job titles, which can make relevant opportunities difficult to find (and diverse in their nature), alongside a lack of recognition of the skills and the role of biocurators both by those performing biocuration activities and in the wider scientific community.

Biocurators that participated in the workshop discussions indicated that one way to address these challenges is to establish stronger links between academia and industry to create a wider and more diverse biocuration community, facilitate knowledge exchange and enable career planning. Another way biocurators can be helped is through appropriate credit, both in terms of citation of the resources they maintain and personal credit through nanopublications, annotation credits and other attributions displayed and linked via their OrCID¹⁷.

Ultimately, we perhaps need a new model of careers in biocuration (especially for those individuals for whom biocuration is their primary role), such as those being developed for Research Software Engineers^18,19 and Data Stewards²⁰. In these settings, centralised teams of research software engineers or data stewards have been created, and this has led to improved knowledge exchange, peer support and career sustainability^19,20. For example, within a University environment, a pool of dedicated biocurators could be created, ready to be deployed where need and experience allows, and supported with local funding, so the biocurators themselves have job security.

We urge funding agencies to recognise the importance of biocuration and to fund both databases and biocuration training activities appropriately. In addition, we recommend that both academia and industry provide clearer career structures for scientists whom biocuration is their primary activity, given their role and impact will only grow. We call on everyone, from the ISB, ELIXIR, academia, industry, and biocurators themselves, to work together to create a more integrated global community of biocurators, to share expertise, training initiatives and best practices.

Conclusion

Biocuration activities have a vital role in the life science data ecosystem, but are often undervalued. These skills, both innate and taught, are highly prized and fundamental not only to the preservation and FAIRness of life science data but are also transferable to other data science and knowledge management roles. The issue we face is one of education. Sometimes scientists performing biocuration activities do not realise the value and rarity of their skills and in many cases researchers and managers do not realise or appreciate the work that biocurators perform. To future-proof this career and the valuable work that biocurators do, organisations, employers and biocurators themselves must act as champions for biocuration and advocate for structural changes. We believe organisations like ELIXIR, ISB and GOBLET have the mandate to actively support the biocurators in their scientific endeavours, work towards raising awareness about the impact that biocuration has within different scientific communities and to convince team leaders of the importance and value of skilled biocuration work. Organisations can also act to facilitate, encourage and support knowledge exchange between biocuration communities. Biocurators themselves have a role to play in championing the work that they do, engaging with the researchers who rely on their expertise and in training the future generation of biocurators. By doing this, organisations and individuals can contribute to the creation of a sustainable and visible biocuration community. We hope the work we have presented here continues this educative process as we continue to define the profession of biocuration.

Data availability