In Methods in molecular biology (Clifton, N.J.)
A capacity to detect the binding profiles of RNA targets for an RNA-binding protein (RBP) under different cellular conditions is essential to understand the functions of the RBP in posttranscriptional regulation. However, the prediction of RBP binding sites in vivo remains challenging. Tools that predict RBP-RNA interactions using sequence and/or predicted structures cannot reflect the exact state of RNA in vivo. PrismNet, which uses both sequences and in vivo RNA structure information from probing experiments, can accurately predict RBP binding under different cellular conditions by deep learning, and can be applied for functional studies of RBPs. Here, we provide a detailed protocol showing how to train a PrismNet model of RBP-RNA interactions for an RBP, and how to apply the model for predictions of the RBP binding under different conditions.
Huang Wenze, Zhang Qiangfeng Cliff
Deep learning, RBP–RNA interaction, RNA structure, RNA-binding protein