Fig. 1: NE score and SCLC molecular subtypes.

a Heatmaps visualizing relative expression of molecular subtype-specific TFs and NE scores. Two heatmaps were generated for each study, with one ordered by complete linkage hierarchical clustering of TFs and the other ordered by NE scores. Gene expression was z-score standardized for each sample. b Pairs plot for NE scores and molecular subtype-specific TFs. Lower left panels are scatter plots showing a pairwise relationship between variables, diagonal panels are density plots showing the distribution of each variable and upper right panels are correlation coefficients from pairwise Pearson correlation. Refer to the top or right of each 5 × 5 matrix for subplot x or y axis variable labels. For example, for each data set, the first subplot on the top row shows the distribution of NE scores from that data set, the scatterplot below it shows the relationship between NE score (x axis value) and ASCL1 expression (y axis value), and the Pearson correlation coefficient between NE score and ASCL1 expression is provided in the second cell of the top row. *p-value < 0.05. Note that in all studies: most samples have positive NE score; under TF-based classification, ASCL1⁺ subtype dominates; NE scores positively correlate with ASCL1 and NEUROD1 but negatively correlate with POU2F3 and YAP1 expression. c H&E staining of two high NE-score SCLC tumor samples showing classic SCLC morphology with dark nuclei, scant cytoplasm, and inconspicuous nucleoli. d ASCL1 IHC staining and H&E staining of a low NE-score SCLC tumor, showing variable morphology at different selected areas, where ASCL1-low areas appear to be more variant-like. e Quantifications of TF expression from IHC staining or microarray profiling, samples are ordered by increasing NE scores. f IHC of ASCL1, NEUROD1, and POU2F3 in two tumors that express both ASCL1 and NEUROD1. Two areas per tumor were selected for showing intratumoral heterogeneity in ASCL1 and NEUROD1 expression patterns.