Study reveals factors that drive efficient, precise editing and hidden DNA repair

Sadie Harley
scientific editor

Robert Egan
associate editor

Researchers at the Max Planck Institute for Evolutionary Anthropology have addressed a longstanding challenge in CRISPR genome editing—accurately predicting guide RNA (gRNA) activity.
Unlike previous prediction tools that were trained using transcribed gRNAs and were therefore prone to transcription biases, the new approach focuses on chemically synthesized gRNAs, which are commonly used in research and are likely to be a preferred format for clinical applications.
In the study in the journal Nature Communications, the researchers developed a simple linear model called the EVA score, which can robustly predict gRNA activity across cell types and datasets.
"Chemical synthesis avoids some of the sequence-related pitfalls of transcribed gRNAs," says Stephan Riesenberg, who led the research.
"This difference, coupled with a more accurate quantification of cellular CRISPR cleavage outcomes, enabled us to build a simple prediction model that generalizes across cell types."
The authors have pre-calculated the EVA scores for all gRNAs in the human and mouse genomes, and these can be easily accessed via online genome browser tracks.
In addition to gRNA activity prediction, the study pioneers a new model for homology-directed repair (HDR) efficiency, a helpful advancement for precise genome editing tasks, such as correcting point mutations. The authors identify sequence features that significantly influence HDR success, such as donor misfolding and type of nucleotide change, enabling HDR efficiencies of up to 78% under optimal conditions.
Beyond editing efficiency, the study also addresses safety concerns critical to therapeutic applications. By examining CRISPR-Cas9 and Cas12a editing outcomes, the researchers uncover a high prevalence of cryptic repair events—genomic changes not detected by standard PCR-based assays, which include large deletions and perfectly repaired breaks indistinguishable from unmodified DNA.
Importantly, the study finds that Cas12a tends to favor perfect repair over large-scale damage, making it potentially safer than Cas9 for therapeutic use. The structural difference in DNA cleavage—Cas12a generates staggered cuts with sticky ends—may explain the divergence in repair outcomes.
The findings have important implications for CRISPR-based research and therapies, offering tools to design more effective and safer genome editing strategies.
More information: Stephan Riesenberg et al, Robust prediction of synthetic gRNA activity and cryptic DNA repair by disentangling cellular CRISPR cleavage outcomes, Nature Communications (2025).
Journal information: Nature Communications
Provided by Max Planck Society