Fig. 1: Molecular subtypes of breast cancer are highly variable across publicly available datasets.
From: Breast cancer gene expression datasets do not reflect the disease at the population level

Distributions of PAM50 intrinsic subtypes assigned by the geneFu package across 70 datasets. The median proportion of luminal A tumors was 25% and luminal B was 31%. Some datasets were dominated by specific subtypes and some were completely lacking in normal-like breast tumors.