Implementations

MicrobiomeAnalyst is developed based on three open source programming languages - Java, R and JavaScript. In particular, the web framework was developed based on the Java Server Faces technology using the Primefaces component library. The metabolic network and 3D interactive visualization is implemented based on the jquery, the sigma.js and the CanvasXpress JavaScript libraries. The backend statistical computation is implemented using R scripts and packages from the Bioconductor project. In particular, the following R packages are used:

• phyloseq: general data manipulation and statistical analysis of metagenomic data;
• vegan: for rarefaction curve analysis and ordination analysis;
• metagenomeSeq, edgeR & DESeq2: statistical comparison of marker genes and shotgun data;
• globaltest: enrichment analysis of metabolic pathways based on KO abundance matrix;
• ggplot2 & ggfortify: data visualization;
• Tax4Fun: prediction of metabolic potentials based on SILVA annotated data;
• RJSONIO: to convert data results to a JSON object for visualization;
• reshape: for data manipulation and processing;
• igraph: for network creation and layout.

MicrobiomeAnalyst is hosted on a powerful server running Ubuntu OS 12.04.2 LTS with 30GB RAM and 8 CPU cores (2.4 GHz each). Please note, the client-side data visualization requires a modern browser that supports HTML5 canvas and JavaScript. MicrobiomeAnalyst has been tested under Google Chrome (5.0+), Firefox (3.0+), and Internet Explorer (9.0+).

Acknowledgement

The work is supported by the NSERC Discovery Grant and the Startup Fund from the McGill University.