Residual parallel convolutional neural network to decipher RNA folding from RNA sequence
The complex structures of the RNA molecules allow them to show diverse functionalities in a cell. In genomics, predicting the secondary structure of RNA from a given nucleotide sequence is an NP-Hard problem. The traditional in vitro and in vivo experimental techniques are time-consuming and expensive. To address this limitation, machine learning, and deep learning methods can be used with the biological domain knowledge to calculate RNA structures efficiently. The earlier approaches were mainly based on the thermodynamics model with the free-energy minimization concept. However, with increasing sequence lengths, RNA takes more complex folds and does not follow the free energy rules. Whereas within a cell, the length of the RNA varies from a few nucleotides to thousands of nucleotides. Here, we propose a deep learning based RNA secondary structure prediction model, named rpcFold, which can efficiently handle RNA with any length and outperforms the existing models on both within-family and cross-family RNA datasets.
