MDSvis, a Shiny application for interactive visualisation of Multi Dimensional Scaling projections

Abstract

MotivationVisual exploratory data analysis is key for making sense of high dimensional data. In particular, Multi Dimensional Scaling (MDS), a low dimensional representation of data points, is frequently used in early stages of data analysis pipelines, for Quality Control (QC), batch effect detection and hypothesis generation. However, interpreting representations in more than two dimensions necessitates the mental integration of multiple two-dimensional (2D) plots - a process that greatly benefits from an interactive Graphical User Interface (GUI), where such plots can be easily and iteratively customised without the need for programming. ResultsHere, we present MDSvis, a Shiny application for user-interactive customised visualisation of MDS projections. MDSvis is based on the Bioconductor package CytoMDS for calculating MDS coordinates. We describe the functional workflow of the application, and illustrate two different use cases, which are detailed in the package vignettes. The first use case pertains to cytometry data analysis, which was the original motivation for building the application. The second use case demonstrates how to extend the use of MDSvis to MDS visualisation of any sample distance matrix. Availability and implementationMDSvis is implemented in the R programming language and is available in a GitHub repository: github.com/UCLouvain-CBIO/MDSvis, under a GPL-3 license.