BGSSN: Breast Cancer-Associated Genes Prediction Based on Weighted Sample-Specific Networks of Cancer Subtypes

Abstract

Breast cancer exhibits a notable degree of heterogeneity in its occurrence and progression, encompassing diverse clinical patterns and outcomes among patients even with identical clinical pathological stages. Genetic mutations in different subtypes of breast cancer may lead to different types of disease and have different clinical implications. Therefore, molecular typing based on the characteristics of breast cancer heterogeneity and the screening of associated genes for different subtypes of breast cancer may be able to more accurately determine the pathogenic genes of breast cancer. In this paper, we propose a weighted sample-specific network based on breast cancer subtypes to predict associated genes, named BGSSN. To better reflect the individual characteristics of patients and the importance of patient samples in different subtypes, the weight of samples is added when constructing the sample-specific network. The random walk with restart method is then utilized to predict new breast cancer-associated genes within the constructed network. By leveraging this method, the network structure can be effectively explored to identify potential gene candidates.