SpCas9 activity prediction by DeepSpCas9, a deep learning–based model with high generalization performance
Problem
Predicting SpCas9 gene-editing activity at arbitrary target sequences was unreliable, limiting CRISPR-based editing.
Why it matters
Unpredictable SpCas9 activity can cause failed edits in gene editing applications across research and medicine.
What the AI did
A deep learning model was trained on high-throughput measurements of SpCas9-induced indel frequencies across 12,832 target sequence pairs to predict activity.
Results
High generalization performance on independently generated datasets (authors' own and from other groups); trained on 12,832 targets.
Impact
Researchers can predict SpCas9 activity through a publicly available web tool.
Record
| Venue | Science Advances |
|---|---|
| Published | 2019 |
| Work type | article |
| Authors | Hui Kwon Kim, Younggwang Kim, Sungtae Lee, Seonwoo Min, Jung Yoon Bae, Jae Woo Choi, Jinman Park, Dongmin Jung et al. |
| Citations | 237 (OpenAlex, at import time) |
| References | 45 |
| Model or method | DeepSpCas9 |
| Organization | not stated in source |
| Primary source | https://doi.org/10.1126/sciadv.aax9249 |
| Open access | gold |
| Keywords | CRISPR, SpCas9, deep learning, activity prediction, indel frequency |
Importance rating
significant — thousands to millions affected
SpCas9 activity prediction broadly relevant to CRISPR editing in research and medicine; abstract gives dataset size (12,832 targets) but no problem scale.