MCAweb: an interactive graphical tool for Multiresolution Correlation Analysis in single-cell data

We present MCAweb, a graphical tool for visualization of local correlations in single-cell gene expression data. MCAweb is an interactive web-based platform for interactive exploration of multivariate data, based on our previously developed method, MCA: multiresolution correlation analysis. It is useful for the identification of cellular subpopulations with differential correlation motifs corresponding e.g. to expression-dependent gene regulatory networks; for the visual identification of outliers contributing to correlations; and for examining the robustness of apparent correlations. MCAweb provides a customizable user interface with functionality for Pearson, Spearman and partial Pearson correlation analysis, visualization of the correlation network for selected subpopulations, and representations of the distributions for selected interaction pairs and sorting variables. Furthermore, pairs of factors showing strong differential regulation are automatically identified, thus serving as a hypothesis generating tool for subpopulation identification.