A method for selecting complementary differentially expressed genes based on mvauc

Abstract

The present invention proposes a feature selection method based on multivariate AUC, which selects the most complementary gene subset from cancer differential expression data to maximize global classification performance. The present invention first proposes a new angle of AUC calculation based on the possible misclassification set of features; then, for a feature set, determine its common possible misclassification set and calculate the new AUC after each feature combination; the new AUC of a feature is compared with the original The difference in AUC shows the complementary effect of other features in the combined feature set on the classification ability of this feature. Finally, mvAUC is calculated based on the new AUC after feature combination, and the candidate features that maximize the current mvAUC are incrementally selected to join the selected feature subset. The method of the present invention has the advantage of being able to directly evaluate the global class discrimination ability of the selected feature subset, and does not need to calculate redundant information between candidate features and each selected feature in pairs.

Description

technical field [0001] The invention belongs to the technical field of data mining, and in particular relates to a method for selecting complementary differentially expressed genes based on mvAUC. Background technique [0002] In the field of biomedicine, with the rapid development and continuous maturity of next-generation sequencing technology (NGS), the cost of sequencing has been greatly reduced, and data such as cancer gene expression has rapidly accumulated, and the analysis and application of big data based on NGS has grown rapidly. Gene expression datasets typically contain tens of thousands or even hundreds of thousands of genes, and a relatively small number of hundreds to thousands of samples. Among the tens of thousands of genes, only a small number of genes are related to the occurrence of cancer, and the existence of a large number of irrelevant redundant genes will seriously affect the analysis of data and lead to bias. It is therefore increasingly important ...

AI Technical Summary

Problems solved by technology

This may overestimate the feature-to-class recognition ability and the redundancy between features, ignoring the overall mutual cooperation of the selected feature subset and the effect on the global classification performance.

Images