A deep learning approach to automate whole‐genome prediction of diverse epigenomic modifications in plants
Problem
Multiple factors limit the scientific utility of plant epigenomic datasets, making whole-genome prediction of diverse epigenomic modifications difficult.
Why it matters
Epigenetic modifications regulate gene transcription and RNA metabolism, so predicting them unlocks plant biology insights.
What the AI did
Developed SMEP, a deep learning model that predicts six types of epigenomic modifications, validated on rice and maize.
Results
95% accuracy for 6mA; ~80% for 5mC, m6A and three histone modifications (10-fold cross-validation); >95% of heat-shock 6mA peaks predicted; applied to indica rice 93-11 and B73 maize.
Impact
Enables cross-line prediction of epigenomic marks in rice and maize, reducing reliance on experiments.
Record
| Venue | New Phytologist |
|---|---|
| Published | 2021 |
| Work type | article |
| Authors | Yifan Wang, Pingxian Zhang, Weijun Guo, Hanqing Liu, Xiulan Li, Qian Zhang, Zhuoying Du, Guihua Hu et al. |
| Citations | 59 (OpenAlex, at import time) |
| References | 65 |
| Model or method | SMEP (Smart Model for Epigenetics in Plants), deep learning |
| Organization | not stated in source |
| Primary source | https://doi.org/10.1111/nph.17630 |
| Open access | hybrid |
| Keywords | deep learning, epigenomics, plants, SMEP, methylation, histone modification |
Importance rating
significant — thousands to millions affected
Epigenomic prediction for rice and maize could reduce experimental reliance and aid genome engineering; abstract claims a genuinely new predictive capability but states no scale of benefit.