Off-target predictions in CRISPR-Cas9 gene editing using deep learning
Problem
Existing methods for predicting off-target mutations in CRISPR-Cas9 gene editing rely on mismatch scores, cannot scale with growing data, and lack clinical-level precision.
Why it matters
Off-target mutations are a key safety risk for clinical gene editing.
What the AI did
Two deep neural networks (a convolutional and a feedforward network) were designed, trained and benchmarked on the CRISPOR off-target dataset, with additional validation on a GUIDE-seq dataset.
Results
not stated in source (abstract truncated before performance metrics)
Impact
Offers a scalable deep learning approach to off-target prediction, benchmarked on independent datasets.
Record
| Venue | Bioinformatics |
|---|---|
| Published | 2018 |
| Work type | article |
| Authors | Jiecong Lin, Ka‐Chun Wong |
| Citations | 185 (OpenAlex, at import time) |
| References | 46 |
| Model or method | deep convolutional neural network and deep feedforward neural network |
| Organization | not stated in source |
| Primary source | https://doi.org/10.1093/bioinformatics/bty554 |
| Open access | hybrid |
| Keywords | CRISPR-Cas9, off-target prediction, deep learning, CNN, gene editing |
Importance rating
significant — thousands to millions affected
Off-target mutations are a key safety risk for clinical gene editing; abstract explicitly ties precision to clinical-level gene editing, an area with major potential impact.