Inference of gene regulatory network using dimension reduction methods and rotation forest

Inferring gene regulation networks from gene expression data has been one of the mostimportant challenges during the last decade. These networks are used in various fields such as diseasediagnosis, gene therapy, etc. In order to infer gene regulatory networks using computational methods,there are various approaches. However, gene expression data are often noisy and gene regulationnetworks are very sparse. In fact, in these networks, the number of regulators of a target gene is verysmall. Most machine learning methods for predicting regulators of a target gene face a high falsepositive rate. In this article, in order to overcome this problem, we first reduce the number of regulatorygenes using dimension reduction methods. Then, in the next step, the regulatory genes of each gene areidentified using the rotation forest. The evaluation results show that dimension reduction methodsincrease the efficiency of rotation forest. In addition, the t-sne method has a better performance thanother dimension reduction methods such as SVD, PCA, etc.

کلیدواژه ها:

Gene regulatory network ، gene expression data ، rotation forest ، dimension reduction ، machine learning