Extended Data Fig. 6: Statistical models based on function-associated signatures can be used to identify tumors with POLE/D1 functional mutations. | Nature Genetics

Extended Data Fig. 6: Statistical models based on function-associated signatures can be used to identify tumors with POLE/D1 functional mutations.

From: Functional landscapes of POLE and POLD1 mutations in checkpoint blockade-dependent antitumor immunity

Extended Data Fig. 6: Statistical models based on function-associated signatures can be used to identify tumors with POLE/D1 functional mutations.The alternative text for this image may have been generated using AI.

a, TCGA data set summary. Wild-type, tumors are wild-type for POLE/D1; Functional, tumors harboring known POLE/D1 functional mutations; Mutated, tumors with only POLE/D1 VUS; SNV, SNV count by WES sequencing. b, the optimal Youden Index point and corresponding probability cutoff value. c, ICGC/CCLE test set summary. d, Heatmap of the Non-negative least squares (NNLS) extracted COSMIC SBS signatures of false negative predictions. e, Reconstitution accuracy of the non-negative matrix factorization (NMF) extracted signatures on TCGA samples with POLE/D1 functional mutations. f, Cosine similarity of the three NMF extracted mutational signatures to the COSMIC SBS signatures. g-h, Contribution of the three NMF de novo mutational signatures to the samples with known POLE/D1 functional mutations in the training set (g) and test set (h). TMB, SNV count in the exome region by WES sequencing. Functional mutation, whether the sample contain POLE or POLD1 functional mutations. i, Fisher exact test on the MSI status of the true-positive and false negative samples from the training and test sets when MSI status is available. j. Tumor allele frequencies of the POLE/D1 functional mutations from the false negative predictions (FN, N = 5), True-positive predictions (TP, N = 77) with available allele frequencies. P value was calculated with two-sided Wilcoxon Rank Sum Test. k, SNV count distribution of the false positive samples from the WES training set. l, Distribution of the true-positive samples (top) or VUS samples (bottom) that were predicted as samples with functional mutations. The green dash lines indicate cutoff for SNVlow (3.6 SNV/Mb exome), the blue dash lines indicate cutoff for SNVint/hi (10 SNV/Mb exome) and the red dash lines indicate cutoff for SNVhyper (50SNV/Mb exome). m. Unsupervised clustering of the SNVint/hi FP samples from the TCGA training set based on the extracted COSMIC SBS signatures.

Back to article page