RNA secondary structure prediction method based on recurrent neural network

Abstract

The invention discloses an RNA secondary structure prediction method based on a recurrent neural network. The RNA secondary structure prediction method includes the steps: carrying out data preprocessing on an RNA primary sequence data set in a PDB data set; dividing the RNA primary sequence into a long sequence, a medium sequence and a long sequence according to the length; vectorizing the sequence information to obtain feature information expressed in a matrix form, and filling the feature information of the sequence samples unsatisfying the standard by taking the longest sequence information of the long sequence, the medium sequence and the short sequence as the standard to obtain a feature matrix with a fixed dimension; and inputting the feature matrix into an LSTM model established based on a recurrent neural network, and performing RNA secondary structure prediction by using the LSTM model. The RNA secondary structure prediction method based on a recurrent neural network can predict the RNA secondary structure, and the prediction result is relatively accurate, and the implicit features of the RNA sequence can be further mined, and the more accurate RNA secondary structure canbe predicted.

Description

technical field [0001] The invention relates to the field of biological research, in particular to a method for predicting RNA secondary structures based on a recurrent neural network. Background technique [0002] Ribonucleotide molecule RNA, as a macromolecule in organisms, is an important substance existing in organisms. It not only cooperates with deoxyribonucleotide molecules DNA and proteins to maintain the activities of organisms, but also in Plays an important role in DNA and protein synthesis. Studies have found that the study of RNA structure can help us understand the function of RNA molecules more comprehensively, which is beneficial for biological researchers to explore the relationship between RNA, DNA and proteins, so as to understand the function of organisms and understand and treat diseases. [0003] The molecular structure of RNA consists of three parts: primary sequence, secondary structure, and tertiary space structure. The tertiary space structure of ...

AI Technical Summary

Problems solved by technology

However, the RNA molecule itself has the characteristics of difficult to obtain crystals and fast molecular degradation, so it is relatively time-consuming and costly to predict its secondary structure by experimental physics and chemical experiments.

The use of computers based on comparative sequence analysis and minimum free energy methods to predict these spatial structures has improved the efficiency of prediction compared with traditional methods, but for secondary structures with long RNA primary sequence bases, the predicted The time and cost are also greatly increased

And these methods are often limited by the biochemical properties of RNA, and the available features of RNA sequences are too few, and traditional machine learning methods are difficult to play a powerful role in the absence of features

Images