Fig. 1: Skeletal muscle gene expression profiles.

Unsupervised hierarchical clustering of individual patients on normalized, log-transformed RNA-seq gene expression data of the 10,000 most variable genes across all patients (a top) and first two principal components calculated using filtered, normalized, log-transformed RNA-seq gene expression data with samples colored according to BC molecular subtype with group geometric means denoted with larger symbols (a bottom). MDS dissimilarity matrix of overall gene expression data represented in three-dimensions, with each dot representing an individual patient. Regression planes are color-coded based on molecular subtype (b), binary disease status (c), or the null model (d).