Igloo-Plot

IglooPlot is a new method to visualize multivariate data in a 2D chart of multiple quantitative variables where features are represented as anchors on a semicircle and samples projected inside the semicircle using a Hooke's law based algorithm. The tool not only identifies clusters in your data but also allows to infer the features responsible for formation of the clusters. Projected data points can be easily mapped to class label information using a simple metadata file. Web-Igloo plots your multi-dimensional data keeping all the features intact without any feature decomposition (unlike in PCA).

Highlighted features include

• User selectable methods for positioning anchors and selecting normalization options
• Options to filter out non informative features
• BoxPlot summaries of anchor (or feature) distribution
• Interactively visualize sample and feature contributions using bar and bubble plots
• Easy mapping of multilevel class label information using a simple metadata file
• Convex hull prediction for class boundaries with automated coloring
• Simple and flexible search function to simultaneously visualize search outputs in plot as well as metadata table
• Interactively identify differentially abundant features across multiple classes using ANOVA
• Custom color selection for class labels

web server LINK : https://web.rniapps.net/webigloo/

Publications :

• https://academic.oup.com/bioinformatics/article/33/4/615/2624550 (The Igloo plot web tool with advanced features)
• https://www.sciencedirect.com/science/article/pii/S0888754314000056?via%3Dihub (The Igloo plot method + basic implementation)