TIMER: a deep-learning framework for general and species-specific bacterial promoter prediction.
TIMER is the first deep learning-based predictor for both general and species-specific bacterial promoters. It uses DNA sequences as input and employs three Siamese neural networks with attention layers to train and optimize the accurate predictive models for 13 species-specific and general bacterial promoters. Extension cross-validation and independent test results demonstrated that TIMER achieved competitive predictive performance and outperformed state-of-the-art predictors on both general and species-specific promoter prediction.
|