Data availabilityDatasets generated in this study have been deposited in public repositories. Both the AF3 prediction outputs (https://doi.org/10.5281/zenodo.17197986 and https://zenodo.org/records/19555204)79,80 and the targeted amplicon sequencing raw count tables (https://zenodo.org/records/17243125)81 are available at Zenodo. The raw dI-profiling sequencing data have been deposited in the Genome Sequence Archive (GSA) in National Genomics Data Center (NGDC) under accession codes HRA013626 and HRA013767 (BioProject PRJCA046878). RNA-seq data for RNA off-target assessments of A3A and TadA8e were downloaded from the SRA database under accession codes PRJNA938578 and PRJNA811343, respectively. Targeted amplicon sequencing data for A3A variant screening were obtained from the SRA database under accession code PRJNA938162. The ChIP–seq and ATAC–seq data relating to chromatin state analysis for off-target sites were downloaded from the ENCODE (ENCFF993NDR, ENCFF480MMN, ENCFF400CMC, ENCFF449FCR, ENCFF567FDM) and GEO (GSM1816075, GSM317585, GSM1816068, GSM1816079, GSM317589, GSM8091609, GSM3762822) databases. These public datasets are listed in the Supplementary Table 6. In this study, the human reference genome assembly GRCh38 (hg38) was used as the reference genome. Source data are provided with this paper.Code availability