Local biplots for multi-dimensional scaling, with application to the microbiome

TL;DR

This paper introduces local biplots as an extension of principal components biplots to multi-dimensional scaling, providing new interpretative tools for understanding variable importance and data smoothing in microbiome analysis.

Contribution

It defines local biplots as Jacobians of MDS maps, linking them to generalized distances and principal components, with applications to microbiome data interpretation.

Findings

Local biplots reveal variable importance in MDS embeddings.

Different phylogenetically-informed distances imply variable data smoothing.

Application to microbiome data shows interpretability of distance variants.

Abstract

We present local biplots, a an extension of the classic principal components biplot to multi-dimensional scaling. Noticing that principal components biplots have an interpretation as the Jacobian of a map from data space to the principal subspace, we define local biplots as the Jacobian of the analogous map for multi-dimensional scaling. In the process, we show a close relationship between our local biplot axes, generalized Euclidean distances, and generalized principal components. In simulations and real data we show how local biplots can shed light on what variables or combinations of variables are important for the low-dimensional embedding provided by multi-dimensional scaling. They give particular insight into a class of phylogenetically-informed distances commonly used in the analysis of microbiome data, showing that different variants of these distances can be interpreted as implicitly smoothing the data along the phylogenetic tree and that the extent of this smoothing is variable.