Figure 3.

Heat map depicting the relative abundance of discriminatory volatile molecules across samples from either late hypoxia (H₈, dark red), early hypoxia (H₁, light red), early normoxia (N₁, light blue), or late normoxia (N₈, dark blue) after log₁₀-transformation, mean centering, and unit scaling. Molecules (rows) are divided into those that were identified in both the four-class and two-class RF models (top), the two-class model only (middle), or the four-class model only (bottom). Cell colors correspond to relative abundance of molecules across samples, ranging from low abundance (blue) to high abundance (red). Dendrogram was calculated using Euclidean distance as a measure of similarity between samples. “Analytes” correspond to molecules for which putative identifications could not be determined, and are accompanied by an experimentally-determined retention index (RI).