Fig. 4: Proteogenomic features underlying whole-genome doubling (WGD) in NSCLC subtypes.

a Barplot showing WGD fraction in each multiomics subtype from our study and CPTAC NSCLC patients. b Overlap of copy number signatures for the five multiomics subtypes, with the colors indicating the odds ratio from one-sided Fisher’s exact test. The COSMIC v3 signature and etiology for each signature are indicated on the y-axis. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 (one-sided Fisher’s exact test). c Top 10 most significantly enriched copy number variations (CNVs) in Subtype 3 (FDR < 0.1). The y-axis indicates the cytoband and the x-axis shows the log₁₀-scaled FDR from linear regressions comparing Subtype 3 and other samples. d Enriched co-mutations in each subtype in both our and the CPTAC cohorts. EGFR mutation, missense mutation, in-frame deletion, frameshift deletion, and amplifications were counted. For TP53 and CDKN2A, amplifications were excluded, and for SOX2, only amplifications were counted. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 (one-sided Fisher’s exact test). e Protein-level gene set enrichment analysis (GSEA) revealing upregulated and downregulated pathways in Subtype 3 of both cohorts. The x- and y-axis are enrichment scores (ES) from the current study and CPTAC NSCLC data, respectively. Labeled pathways are the top-5 upregulated pathways in Subtype 3. The Molecular Signatures Database (MSigDB) hallmark gene set v7.4 was used for GSEA. f Subtype 3-specific kinase activity scores. The sizes of the points indicate -log₁₀(FDR) from kinase activity estimation. Significantly up- and downregulated kinases are labeled (FDR < 0.05). g Elevated protein expression of XPO1 in Subtype 3 is shown for our study (Subtype 1, n = 55; Subtype 2, n = 45; Subtype 3, n = 52; Subtype 4, n = 43; Subtype 5, n = 34) and CPTAC LSCC (Subtype 1, n = 11; Subtype 2, n = 19; Subtype 3, n = 42; Subtype 4, n = 15; Subtype 5, n = 21). Kruskal-Wallis test was performed to test the differences in expression. WGD status is marked by red dots and the y-axis shows log₂ protein expression levels. For box-plots, middle line, median; box edges, 25^th and 75^th percentiles; whiskers, most extreme points that do not exceed ± 1.5 × IQR. h Sample information (top), drug response curve (middle), and IC50 (bottom) for selinexor (XPO1 inhibitor) for lung organoids highlighting a higher sensitivity in WGD-positive LSCC organoids. Three technical replicates were tested in each organoid sample. For IC50 barplot, dots indicate each replicate, and error bars indicate average ± 1 standard deviation. i WGD-related pathway underlying Subtype 1 LUAD tumors and Subtype 3 LSCC tumors. Significantly upregulated kinases are highlighted with red triangles (FDR < 0.05) and mutations are shown in purple boxes. Kinase activity scores are estimated from phosphoprotein expression. The log₂ fold changes from DE analyses are indicated by the color in each box. For the phosphoproteome, only features with FDR less than 0.1 are displayed.