MRDGC: A Parallel Approach for the Identification of Master Regulators Based on the Differently Expressed Genes and the Regulatory Capacity of Regulators

摘要

Master regulators play a major role in most survival and development process of living organisms. Identifying the master regulators provides great insight into analyzing diseases and designing drugs. Many computational strategies for master regulators identification have been put forward based on the expression profiles and single relations among transcription factors (TFs), microRNAs (miRNAs) and genes. However, the regulatory relations are sparse and most methods can only recognize one of transcription factors or miRNAs with inaccurate parameters. In this article, we present a novel nonparametric parallel approach for the identification of master regulators based on the differentially expressed genes and the regulatory capacity of regulatory factors, called MRDGC. The regulators having significant meaning in Wilcoxon rank-sum test are selected as candidate master regulators in parallel and the master regulators are the TOP k candidate master regulators ranked ascending by statistical values. In the computational experiments, MRDGC achieves good robustness and outperforms most of methods in breast and thyroid cancer. Moreover, a detailed case study on breast cancer demonstrates the ability of MRDGC for uncovering master regulators. MRDGC is implemented in MATLAB and available to download at https://github.com/sunmingming99/MRDGC.