An IscB protein derived from the human gut metagenome (OgeuIscB), a hyper-compact RNA-guided endonuclease encoded by IS200/IS605 transposons, is considered an evolutionary ancestor of Cas9. It consists of only ~500 amino acids, making it approximately...
An IscB protein derived from the human gut metagenome (OgeuIscB), a hyper-compact RNA-guided endonuclease encoded by IS200/IS605 transposons, is considered an evolutionary ancestor of Cas9. It consists of only ~500 amino acids, making it approximately three times smaller than SpCas9. Unlike Cas9, IscB lacks the REC domain and instead relies on an associated noncoding RNA known as OMEGA RNA (ωRNA) for target DNA recognition. The wild-type IscB system exhibits limited genome editing activity in human cells; therefore, several IscB and ωRNA variants have been developed through mutational scanning and rational design. In this study, we first evaluated the editing efficiencies of 30 combinations, comprising five IscB proteins and six ωRNA scaffolds, at 805 target sequences. Next, we thoroughly characterized the enIscB–ωRNA* system, which exhibited the highest editing efficiency among the 30 combinations, using a library comprising 23,195 ωRNA*–target sequence pairs. The resulting large-scale dataset enabled us to determine the target-adjacent motif (TAM) preference of enIscB and to profile on-target and off-target editing activities across a broad range of target sites. In addition, leveraging this dataset, we developed a deep learning model capable of predicting IscB activity and designing optimized ωRNA* sequences for given target sites. Altogether, this study provides a comprehensive framework for optimizing genome editing using a hyper-compact IscB system and offers valuable insights for its future application in therapeutic and precision genome engineering.