Figure 1
From: Bayesian probit regression model for the diagnosis of pulmonary fibrosis: proof-of-principle
Unsupervised cluster of the complete dataset (training cohort). Samples include normal lung (black) and IPF (brown). IPF is divided into biopsy (red) and explant (green). Samples are also identified by their lobe of origin: upper lobe (orange), lower lobe (blue) or unknown lobe (yellow). (A) Unsupervised hierarchical clustering of all samples based on gene expression profiles. Samples include 6 normals (Normal_[A through F]) and 17 samples of IPF (6 upper/lower lobe pairs and 5 singletons) of which 8 are biopsies (Biopsy_[3-digit sample ID][U = upper or L = lower] and 9 are explants (Explant_[3-digit sample ID][U = upper or L = lower]). (B) Enlargement of the dendrogram, sample names and color key from Figure 1A. (C) Samples are plotted according to expression of the first two Principal Components. [Key: singletons = colored square; all other shapes represent lobar pairs].