JMB-HEADER RAS-JOURNALS EIMB Pleiades Publishing

RUS

             

ENG

YearIMPACT-FACTOR
2022  1,200
2021  1,540
2020  1,374
2019  1,023
2018  0,932
2017  0,977
2016  0,799
2015  0,662
2014  0,740
2013  0,739
2012  0,637
2011  0,658
2010  0,654
2009  0,570
2008  0,849
2007  0,805
2006  0,330
2005  0,435
2004  0,623
2003  0,567
2002  0,641
2001  0,490
2000  0,477
1999  0,762
1998  0,785
1997  0,507
1996  0,518
1995  0,502
Vol 53(2019) N 1 p. 118-126; DOI 10.1134/S0026893319010175 Full Text

Y.Z. Wang1, J. Li1, S. Zhang1, B. Huang1, G. Yao1, J. Zhang1*

An RNA Scoring Function for Tertiary Structure Prediction Based on Multi-layer Neural Networks

1School of Physics, Collaborative Innovation Center of Advanced Micro structures, and National Laboratory of Solid State Microstructures, Nanjing University, Nanjing, 210093 China

*jzhang@nju.edu.cn
Received - 2018-01-10; Revised - 2018-04-26; Accepted - 2018-05-17

A good scoring function is necessary for ab inito prediction of RNA tertiary structures. In this study, we explored the power of a machine learning based approach as a scoring function. Compared with the traditional scoring functions, the present approach is more flexible in incorporating different kinds of features; it is also free of the difficult problem of choosing the reference state. Two multi-layer neural networks were constructed and trained. They took RNA a structural candidate as input and then output its likeness score that evaluates the likeness of the candidate to the native structure. The first network was working at the coarse-grained level of RNA structures, while the second at the all-atom level. We also built an RNA database and split it into the training, validation, and testing sets, containing 322, 70, and 70 RNAs, respectively. Each RNA was accompanied with 300 decoys generated by high-temperature molecular dynamics simulations. The networks were trained on the training set and then optimized with an early-stop strategy, based on the loss of the validation set. We then tested the performance of the networks on the testing set. The results were found to be consistently better than a recent knowledge-based all-atom potential.

RNA structure prediction, scoring function, machine learning, neural network



JMB-FOOTER RAS-JOURNALS