Analyses of potential off-target editing by targeted amplicon deep sequencing. Genomic DNA from 3 mice with the highest on-target editing and 2 untreated control mice was used. (A) Off-target prime editing at the top 20 potential sites experimentally identified by CIRCLE-seq (left panels), and the top 5 sites predicted in silico by Cas-OFFinder (right panels). After sequencing, off-target prime editing was calculated based on the percentage of reads with G/T/C > A conversion at position +4 (corresponding to the sickle mutation), counting the predicted nicking site as position +1. If position +4 is already an A in the wild-type allele, a calculation was performed based on the percentage of G/T/C > A conversion at position +5 (corresponding to the silent PAM mutation). Each symbol represents an individual animal. (B) Percentage of reads with indels. The indel levels were analyzed using a Cas-Analyzer. Genomic DNA samples were the same as those in panel A.