Qualitative Biclustering with Bioconductor Package rqubic

Biclustering has been suggested and found very useful to discover gene regulation patterns from gene expression microarrays. Several quantitative algorithms, among others CC and BIMAX, have been implemented in R, mainly by the biclust package. To our best knowledge, there have been so far no qualitative biclustering methods implemented. Therefore we introduce rqubic, a Bioconductor package implementing the qualitative biclustering (QUBIC) algorithm. Compared to quantitative alternatives, this algorithm is less sensitive to outliers and heavy tails of microarray data. In addition, it is straightforward to plug other discretized data types in the algorithm, for example differentially expressed transcripts in NGS experiments, for which several R packages (e.g. edgeR and DESeq) have been available. The vignette introduces the functionalities provided by the rqubic package. And we demonstrate the usage of the package by implementing a biclustering software pipeline.