Objective: In order to improve the survival rate of breast cancer patients and improve the clinical treatment of patients, it is necessary to study the pathogenic genes of breast cancer from a molecular mechanism. Methods: Firstly, the differential expression of 113 normal tissues and 1109 cancer tissues was analyzed. Then, a group of complementary genes was grouped in a conditional joint analysis method for differentially expressed genes, and a set of gene fitting prognostic models was selected using stepwise cox regression. Results: The six genes VWCE, SPDYC, CRYBG3, DEFB1, SEL1L2, and NMNAT2 have a harmful effect on survival rate. The four genes AMZ1, GJB2, CXCL2, and ALDOC are beneficial to survival rate.The final prognostic model of the 10 genes can significantly divide the sample into high-risk group and low-risk group, and predict the 5-year and 10-year survival rates of breast cancer patients, and the time-dependent AUC values are both Up to 0.7 or more. Conclusion: This method can take advantage of the correlation between genes to reduce dimensionality of high-dimensional data and eliminate the problem of collinearity between genes. The prognostic model of these 10 genes can predict the clinical outcome of patients provide help.