Association Plots: visualizing cluster-specific associations in high-dimensional correspondence analysis biplots

Abstract In molecular biology, just as in many other fields of science, data often come the form matrices or contingency tables with observations (rows) for a set variables (columns). While projection methods like principal component analysis correspondence (CA) can be applied obtaining an overview such data, cases where matrix is very large associated loss information upon into two three dimensions may dramatic. However, when grouped clusters, this opens up new angle on data. We focus question which are to cluster and distinguish it from clusters. CA employs geometry geared towards answering question. exploit feature order introduce Association Plots visualizing cluster-specific complex Regardless dimensionality two-dimensional depict variables. demonstrate our method small sets then use study challenging genomic comprising >10,000 samples. show that clearly highlight those characterise