How far are we from the era of big data in transcriptomics? Lessons from the bacterial data in GEO.

Abstract

The Gene Expression Omnibus (GEO) is the largest functional genomics repository, including ~5 million entries related to the main transcriptomic technologies: microarrays and RNA-seq. This amount of data has the potential to be reused in large-scale meta-analysis, such as those in bacterial systems biology, where the landscape of biological conditions is wider and more diverse than any individual experiment alone. Notwithstanding the accelerated growth in RNA-seq experiments, microarray still accounts for ~48% of bacterial transcriptomic entries in GEO, highlighting the need to revalue this data. Therefore, in this work, we assess the current state of bacterial microarray and RNA-seq data and metadata. We report diverse inconsistencies in both the GEO metadata documentation and community usage, limiting the automated access to biological context essential for high-throughput analysis interpretation. Additionally, while access to and analysis of RNA-seq data are topics widely discussed by the community, microarray data processing and normalization present challenges that need to be addressed for the proper data integration into large-scale reanalysis. Thus, we delve into the availability and processability of bacterial microarray data in GEO, showing a complex panorama where the lack of standard formats limits our reusability potential to at least 44% of the ~45 000 microarray entries. We conclude that GEO transcriptomic data and metadata should be viewed as valuable resources that require ongoing revision and maintenance. Finally, we propose a series of guidelines to enhance the Findability, Accessibility, Interoperability, and Reusability of GEO, thereby taking a step forward into the era of big data.