Genomics data curation roles, skills and perception of data quality

Abstract

Compared to a decade ago, genomics scientists, driven by technical changes and availability of massive genomics data, are performing a wider plurality of curation roles, including end user, curator, and dual-role user. Scientists with different curation roles (including that of end user) may focus on different data quality aspects and skill requirements in a community curation environment. This study examines how genomics scientists' perceived priorities for data quality and data quality skills differ when assuming different roles played in genomics data curation work. The analysis of survey data collected from 147 genomics scientists found that curators of genomics data valued quality criteria that can be assessed through direct examination of the data more highly, while end users placed a high value on the quality criteria that can be assessed indirectly, such as believability. With regard to data quality skills, curators appeared to care more about understanding user's requirements and specific data management skills than end users, while end users valued the skills needed to deal with information overload more highly — those needed to identify useful, relevant information from large amounts of data. Scientists with different curation roles, given common curation tasks with the same skill requirements, prioritized different data quality criteria. The data quality, skill priorities, and tradeoffs identified by this study can inform the development of effective data curation mandates and policies, data quality assurance planning and training, and the design of curation role specific tool dashboards and visualization interfaces for genomics data.