Screening method for radiosensitivity marker genes with balanced clinical confounding factors

Abstract

The present invention divides malignant tumor patient data into clinical factors and pre-radiotherapy gene expression profile data, namely gene sets; constructs a regression model of clinical factors based on whether the clinical factors are used or not, and calculates the clinical factors PS of each sample; Two independent samples t-test test was performed as the group, and the differentially expressed gene subset with P<0.1 was screened out; PS was included as a fixed independent variable, and the expression level of one of the remaining genes in the gene subset was included as an independent variable each time. For the regression model with radiosensitivity or tolerance as the outcome, the area under the curve AUC was calculated, and the gene expression with the largest AUC was selected to be included in the regression model, until the AUC of the regression model reached the maximum and the cycle ended to form a multiple regression model; genes included in the multiple regression model It constitutes a radiosensitivity marker gene; based on the multiple regression model, the dependent variable or its inverse is obtained as the radiotherapy sensitivity score of the patient, and a certain cutoff value is used to predict whether the patient is radiosensitive or not.

Description

technical field [0001] The invention relates to a method for screening radiotherapy sensitivity marker genes based on gene expression profiles of clinical samples, and in particular proposes a screening method for balancing clinical confounding factors. Background technique [0002] Surgery, chemotherapy, and radiotherapy are the three traditional methods for the treatment of malignant tumors. With the development of omics technology, the treatment of malignant tumors has gradually entered the era of precision medicine. The development of radiation oncology is currently at such an inflection point that the eyes of molecular biology and genetics can be more accurately targeted. In recent years, individualized precision radiotherapy models based on genomics models have been gradually developed and applied. [0003] At present, most of the methods to find the main genes that affect radiosensitivity are through cell in vitro experiments combined with gene expression profiling....

AI Technical Summary

Problems solved by technology

In particular, the treatment of tumors is often comprehensive treatment, and radiotherapy is often combined with chemotherapy or molecular targeted drug therapy, so other treatment factors also greatly interfere with the evaluation of radiotherapy efficacy

Images