Confounding factors in assessing the enriched expression of somatic mutant alleles in bulk tumor samples

Evaluation of the AEV model using somatic truncation indels in TCGA samples. (A) Comparison of elevated AEV indels (AEV > 1.2 with FDR < 0.01) in tumor-suppressor genes (TSGs) and nondriver genes. Scatter plots showing VAF^DNA (x-axis) and VAF^RNA (y-axis) for indels in TSGs and nondrivers with elevated AEVs highlighted in red dots. Nondriver indels were down-sampled to match the TSG indel size for visualization. (B) The prevalence of elevated AEV in TSGs and nondrivers was compared by Fisher's exact test using the full data set, which contained 4450 TSG and 69,840 nondriver indels, respectively (left). The involvement of LOH at indel loci was also compared by Fisher's exact test (right). (C,D) Evaluation of expression difference in tumor versus normal on elevated AEV is shown for differential expression (C) and for global upregulation in transcription activity (hypertranscription; D). For measuring differential expressions in C, relative expression per copy (e) was surrogated by dividing each tumor sample's TPM with the median TPM in tissue-matched normal samples adjusted for copy number. Box plots compare the distribution of the TMP ratio for samples harboring elevated AEV indels (elevated) versus all indels in the corresponding gene category (overall) with P-values by Mann–Whitney U test. The expression ratio plotted in D was based on the RNA amount ratio estimated by Zatzman et al. (2022). (E,F) The effect of tumor purity on elevated AEV. (E) Comparison of tumor-purity distribution in samples harboring elevated indels (elevated) versus all indel-containing samples (overall) by Mann–Whitney U test. (F) Correlation of elevated AEV prevalence (y-axis) in a tumor type with the median purity of the tumor type (x-axis) by linear regression. (PAAD) Pancreatic adenocarcinoma; (HNSC) head and neck squamous cell carcinoma; (LUAD) lung adenocarcinoma; (LAML) acute myeloid leukemia; (GBM) glioblastoma multiforme; (LGG) lower-grade glioma; (ACC) adrenocortical carcinoma; (UVM) uveal melanoma. (G,H) The effect of NMD on elevated AEV is shown in G for the comparison of NMD features by χ² test (G) and in for comparison of the prevalence of LOH by Fisher's exact test stratified by NMD efficiency (d > 0 vs. d ≈ 0; H).