RNA secondary structure prediction using an ensemble of two-dimensional deep neural networks and transfer learning

By Jaswinder Singh, Jack Hanson, Kuldip K. Paliwal, Yaoqi Zhou. 2019

348 citations recorded by the source corpus.

The majority of our human genome transcribes into noncoding RNAs with unknown structures and functions. Obtaining functional clues for noncoding RNAs requires accurate base-pairing or secondary-structure prediction. However, the performance of such predictions by current folding-based algorithms has been stagnated for more than a decade. Here, we propose the use of deep contextual learning for base-pair prediction including those noncanonical and non-nested (pseudoknot) base pairs stabilized by tertiary interactions. Since only <\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$<$$\end{document}250 nonredundant, high-resolution RNA structures are available for model training, we utilize transfer learning from a model initially trained with a recent high-quality bpRNA dataset of >\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$> $$\end{document}10,000 nonredundant RNAs made available through comparative analysis. The resulting method achieves large, statistically significant improvement in predicting all base pairs, noncanonical and non-nested base pairs in particular. The proposed method (SPOT-RNA), with a freely available server and standalone software, should be useful for improving RNA structure modeling, sequence alignment, and functional annotations. The limited availability of high-resolution 3D RNA structures for model training limits RNA secondary structure prediction. Here, the authors overcome this challenge by pre-training a DNN on a large set of predicted RNA structures and using transfer learning with high-resolution structures.

Subfields: Molecular Biology

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Molecular Biology

1. Yang Zhang
2. Chengxin Zhang
3. Yang Li
4. Robin Pearce
5. Yaoqi Zhou
6. Jian Zhan
7. Peng Xiong
8. Pengyi Yang

Citations in the corpus

• Interpretable RNA Foundation Model from Unannotated Data for Highly Accurate RNA Structure and Function Predictions (2022, methodology): “We fine-tune our secondary structure predictor on the PDB dataset [28],”
• Ensemble deep learning in bioinformatics (2020, methodology): “The replacement of CNN with ResNet 73 as well as the addition of residual connections between GRU and CNN 74 were also explored to facilitate feature propagation for improved modelling of long-range dependencies between amino acids.”
• Integrating end-to-end learning with deep geometrical potentials for ab initio RNA structure prediction (2022, positive): “For example, SPOT-RNA 11 , MXfold2 12 and Ufold 13 utilize convolutional neural networks (CNNs) or recurrent neural networks (RNNs) to improve the accuracy of secondary structure (SS) prediction for RNAs.”
• Pairing a high-resolution statistical potential with a nucleobase-centric sampling algorithm for improving RNA model refinement (2021, background): “In other words, this algorithm will become more useful as secondary and tertiary base pairs are increasingly more accurately predicted by employing co-variational analysis of RNA homologous sequences 39 – 42 and deep learning techniques 43,44 .”