System for predicting influence of amino acid variation on protein structure stability, and method thereof

Abstract

The invention discloses a system for predicting the influence of amino acid variation on protein structure stability, and a method thereof. The system consists of an amino acid variation information input module, an amino acid variation site attribute calculation module, a protein sequence attribute calculation module, a prediction stability change module and a prediction result output module. The method comprises the following steps that: inputting and obtaining variation information; extracting AAindex attribute features, and calculating the micro-electrical physical and chemical attribute features of amino acid; calculating the conservative property and the protein attribute of a protein sequence corresponding to the amino acid variation; adopting a two-layer three-classification random forest algorithm to calculate the influence of the amino acid variation on the protein stability; and storing and outputting a prediction result. By use of the system and the method, on the basis of the amino acid variation provided by a user and the corresponding protein sequence, a situation that the structure stability of protein where the amino acid is positioned is raised and lowered or is invariant due to the amino acid variation is accurately predicted, a corresponding probability is also accurately predicted, and the result is stored and sent to the user to be stored.

Description

technical field [0001] The invention belongs to the technical field of biomedical data analysis, and in particular relates to a system and method for predicting the influence of amino acid variation on protein structure stability. Background technique [0002] An important index to predict the effect of amino acid variation on protein stability is the free energy change value ddG of the wild-type protein and the protein after mutation. At present, the existing prediction methods are divided into two types: one is to calculate directly based on energy and use physical formulas, but due to the unclear physical structure of proteins, the calculation results are not accurate and the generalization is weak; [0003] The other is to use machine learning methods to predict based on existing experimental data, but this method will have the following problems: [0004] (1) The accuracy is poor. Currently, there are many data errors and omissions in the general experimental database ...

AI Technical Summary

Problems solved by technology

[0004] (1) The accuracy is poor. Currently, there are many data errors and omissions in the general experimental database Protherm, which leads to poor quality of the training data set and seriously affects the accuracy of the prediction results;

[0005] (2) Poor generalization. This method uses a large number of protein structure-related input attributes, but it cannot be predicted when the protein structure is unknown.

[0006] (3) Poor practicability. This method lacks a system that supports single and batch input and can classify prediction results into three categories (variation leads to increased, decreased, and unchanged protein stability)

Images