Prediction of base editor off-targets by deep learning
Problem
Base editors frequently induce unwanted Cas9-dependent off-target DNA mutations, and researchers lacked tools to predict where these off-targets occur.
Why it matters
Unpredicted off-target mutations can compromise the safety of base-editing applications.
What the AI did
Deep learning models (ABEdeepoff, CBEdeepoff) were trained on tens of thousands of measured off-target efficiencies to predict off-target sites, made freely accessible via a web server.
Results
Spearman correlations of 0.710 to 0.859 predicting off-targets at endogenous loci; training datasets of 54,663 (ABE) and 55,727 (CBE) off-targets
Impact
A free web server lets researchers predict base editor off-targets to help minimise off-target effects.
Record
| Venue | Nature Communications |
|---|---|
| Published | 2023 |
| Work type | article |
| Authors | Chengdong Zhang, Yuan Yang, Tao Qi, Yuening Zhang, Linghui Hou, Jingjing Wei, Jingcheng Yang, Leming M. Shi et al. |
| Citations | 39 (OpenAlex, at import time) |
| References | 52 |
| Model or method | ABEdeepoff and CBEdeepoff (deep learning models) |
| Organization | not stated in source |
| Primary source | https://doi.org/10.1038/s41467-023-41004-3 |
| Open access | gold |
| Keywords | base editing, off-target prediction, deep learning, Cas9, web server |
Importance rating
significant — thousands to millions affected
Base editors frequently induce unwanted off-target mutations; predictive models with a free web server support safety of base-editing applications across the field.