Abstract—Identifying gene signatures that are associated with the estrogen receptor based breast cancer samples is a challenging problem that has significant implications in breast cancer diagnosis and treatment. Various existing approaches for identifying gene signatures have been developed but are not able to achieve the satisfactory results because of their several limitations. Subnetwork-based approaches have shown to be a robust classification method that uses interaction datasets such as protein-protein interaction datasets. It has been reported that these interaction datasets contain many irrelevant interactions that have no biological meaning associated with them, and thus it is essential to filter out those interactions which can improve the classification results. In this paper, we therefore, proposed a hub-based reliable gene expression algorithm (HRGE) that effectively extracts the significant biologically-relevant interactions and uses hub-gene topology to generate the subnetwork based gene signatures for ER+ and ER- breast cancer subtypes. The proposed approach shows the superior classification accuracy amongst the other existing classifiers, in the validation dataset.

Index Terms—Breast cancer diagnosis, estrogen-receptor, gene signature, hub-gene.

Ashish Saini is with Deakin University, Victoria, Australia (e-mail: asain@deakin.edu.au).

Cite: Ashish Saini, Jingyu Hou, and Wanlei Zhou, "Hub-Based Reliable Gene Expression Algorithm to Classify ER+ and ER- Breast Cancer Subtypes," International Journal of Bioscience, Biochemistry and Bioinformatics vol. 3, no. 1, pp.  20-26, 2013.