Enhancing CRISPR-Cas9 gRNA efficiency prediction by data integration and deep learning
Problem
Designing CRISPR gRNAs requires accurate on-target efficiency predictions, which existing tools could not provide well enough.
Why it matters
Accurate gRNA efficiency prediction underpins reliable CRISPR applications in research and therapy.
What the AI did
Trained a deep learning model, CRISPRon, on 23,902 gRNAs, integrating newly generated activity data for 10,592 SpCas9 gRNAs with published data; released standalone software and a webserver.
Results
Significantly higher prediction performance than existing tools on four test datasets not overlapping with those tools' training data.
Impact
CRISPRon provides more accurate gRNA efficiency predictions than existing tools, publicly available as software and webserver.
Record
| Venue | Nature Communications |
|---|---|
| Published | 2021 |
| Work type | article |
| Authors | Xi Xiang, Giulia Ilaria Corsi, Christian Anthon, Kunli Qu, Xiaoguang Pan, Xue Liang, Peng Han, Zhanying Dong et al. |
| Citations | 206 (OpenAlex, at import time) |
| References | 44 |
| Model or method | CRISPRon (deep learning model) |
| Organization | not stated in source |
| Primary source | https://doi.org/10.1038/s41467-021-23576-0 |
| Open access | gold |
| Keywords | CRISPR, gRNA efficiency, deep learning, CRISPRon, genome editing |
Importance rating
significant — thousands to millions affected
Accurate gRNA efficiency prediction underpins reliable CRISPR applications across research and therapy; validated on 23,902 gRNAs with better performance than existing tools, a genuinely enabling capability.