Random forest-based sucrose transporter identification method

Abstract

The invention relates to a random forest-based sucrose transporter identification method, which comprises the following steps of: firstly, acquiring initial data from a protein database, preprocessing the initial data, deleting sequences containing non-standard letters and sequences with too short lengths, and deleting sequences with the similarity greater than 60%; extracting different features according to physicochemical properties and evolutionary information of the protein, and taking each feature and the combined feature as feature input; then, due to the fact that the difference between the sample numbers of the positive example and the counter example is large, performing oversampling on the data set; and finally, under ten-fold cross validation, performing feature training on the oversampled training set by using a random forest, a support vector machine, stochastic gradient descent and naive Bayes, performing testing by using a test set, and analyzing a result. According to the method, a k-separated-bigrams-PSSM and random forest combined method is used, and a Borderline-SMOTE algorithm is introduced, so that the problem of data imbalance is solved, and the identification accuracy of the sucrose transporter is effectively improved.

Description

Technical field: [0001] The invention relates to the field of protein classification, in particular to a method for identifying sucrose transporters based on random forests. Background technique: [0002] Sucrose transporter is a transmembrane protein widely present in plants. It plays an important physiological function in the process of sucrose transport and sucrose-specific signal induction, and has great research value on plant growth and yield. Traditional biochemical techniques require the preparation of a large number of experimental supplies, and the detection process is complex, time-consuming and labor-intensive. Currently, more and more researchers use machine learning methods to classify proteins. [0003] First of all, classifying proteins requires obtaining a benchmark data set, then the original data set needs to be preprocessed to ensure reliability. Next, feature extraction is performed on the sequence, mainly including amino acid composition, physical and...

AI Technical Summary

Problems solved by technology

[0005] The purpose of the present invention is to classify a new protein - sucrose transporter, and propose a method based on machine learning, especially a method based on random forest and k-separated-bigrams-PSSM, to solve Problems in the classification of sucrose transporters

Images