一种基于深度学习算法预测pre-mRNA剪切的方法及系统
By combining convolutional neural networks with self-attention mechanisms from deep learning algorithms, a neural network model was constructed, which solved the problem of insufficient prediction ability of existing RNA splicing methods for recessive non-classical splicing sites. This achieved high accuracy and rapid prediction, promoting the development of ASO drugs and personalized treatment.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- TONGJI UNIV
- Filing Date
- 2024-08-01
- Publication Date
- 2026-07-17
AI Technical Summary
Existing RNA splicing methods mainly rely on known splicing sites as training sets, resulting in weak prediction ability for recessive non-classical splicing sites, unsatisfactory accuracy, and slow speed. Traditional machine learning models are not flexible enough in recognizing complex patterns and have difficulty handling large numbers of samples.
A neural network model combining convolutional neural networks based on deep learning algorithms and self-attention mechanism was adopted. Pre-mRNA sequence data were obtained from public databases and RNA-seq sequencing datasets, preprocessed and labeled, and a neural network model was constructed. Three-class classification prediction was performed to extract the probabilities of donor/recipient splicing sites and non-splicing sites.
It improves the accuracy and speed of RNA splicing site identification, enhances the feature input for non-classical splicing sites, and enables the trained model to quickly assess the impact of gene variations on RNA splicing, thus promoting the development and personalization of ASO drugs and reducing R&D costs.
Smart Images
Figure CN119181423B_ABST