2020-12-15
Presentations at EuroBioc2020
At the European Bioconducutor Meeting EuroBioc2020, Dec 14-18, 2020, we will give a short talk, birds of a feather session, and a poster on the emerging R ecosystem for microbiome research.
Links to the material:
Dedicated class structures are fundamental for the R/Bioconductor ecosystem. In microbiome research, the phyloseq class has become a commonly accepted standard. However, recent developments in Bioconductor classes and growing sample sizes have opened up new opportunities and requirements to extend this popular format in order to address emerging research needs. Extended support is needed for instance for hierarchically structured data, linking with other data types, and for general performance optimization.
We propose MicrobiomeExperiment as a novel class structure for microbiome data. This extends the TreeSummarizedExperiment and SingleCellExperiment classes, which provide thoroughly tested tools for hierarchical data, spanning both sample and feature spaces. The new class adds support for additional feature information that is specifically relevant for microbiome experiments; it allows a seamless incorporation of more detailed sequence information based on existing classes such as the DNAStringSet(List). The MicrobiomeExperiment class inherits support for sparse matrices and multiple assays while providing improvements in speed and memory compared to the currently available solutions for microbiome data. Importantly, it provides conversion methods from the widely used phyloseq class as well as other raw data types, thus enabling seamless conversion.
By linking microbiome data more tightly to other already established and strongly supported Bioconductor classes, we hope to reduce overlapping development efforts, improve the interoperability of available tools, and simplify the addition and ensure the long-term sustainability of the new tools. MicrobiomeExperiment is aimed at solving shortcomings in the current microbiome R ecosystem and it has the potential to become a widely accepted standard for microbiome data in the R/Bioconductor ecosystem. The class structure and the associated package ecosystem are now under active development as a joint effort of multiple research teams, and feedback from the community will be highly valued.