Extended Data Fig. 2: Optimization of PEER-seq and comparison with pegRNA-based analysis.

a, Correction of positional biases. To correct positional biases in LFCs, we employed LOWESS regression by using synonymous SNVs, which were assumed to have no functional effect. The LOWESS regression curves are shown as orange lines. The overall depletion of nonsense SNVs was more distinct after the adjustment. b, ROC-AUC analysis to determine the effect of positional bias correction for sets of nonsense (the number of SNVs n = 27) vs. synonymous SNVs (n = 117) in exon 2 of RPL15. Pre, ROC-AUC before adjustment; Post, ROC-AUC after adjustment. c, Correlation between PEER-seq LFC values from two biological replicates. The Pearson correlation coefficient (r) is shown. The number of SNVs n = 511. d,e, Heatmaps showing adjusted LFCs of 516 (=172 × 3) SNVs (d) and 336 protein variants (e) generated by prime editing in exon 2 of RPL15. SNVs (d) and protein variants (e) with P-values of two-sided Fisher’s exact test greater than 0.05 or odds ratios lower than 3 were excluded from the analysis and are shown with a white background. SNVs (d) and protein variants (e) for which no pegRNAs were designed are shown as gray boxes. The numbers at the bottom of each heatmap represent the location in the RPL15 coding sequence (d) and in the RPL15 amino acid sequence (e). At each position, the nucleotide (d) and amino acid (e; WT, wild-type; top) in the reference sequences are shown. f, Correlation between adjusted LFC values of SNVs determined by pegRNA abundance-based analysis from two biological replicates. The Pearson correlation coefficient (r) is shown. The number of SNVs n = 511. g, Kernel density estimation plots of adjusted LFCs of SNVs in RPL15 determined by pegRNA abudance-based analysis as a function of the SNV category. For each category, the number and percentage of SNVs with adjusted LFC values lower than a cutoff value (the gray dashed vertical line), representing the 5th percentile of the adjusted LFC values of synonynous mutations, are shown. h, Correlation between adjusted LFC values of SNVs calculated from PEER-seq evaluations and those from pegRNA abundance-based analysis. The Pearson correlation coefficient (r) is shown. The number of SNVs n = 511.