Predicting Disease-related Genes Using Biomedical Literature Based on GloVe Word Embedding

Abstract

Identifying disease-related genes is essential for understanding disease mechanisms and treating patients. Because wet-lab experiments are time-consuming and expensive, studies that use text mining have been increasing. In this paper, we propose a method to predict novel disease-related genes based on GloVe. Existing word embedding-based methods do not consider a characteristic of word embedding, so it is difficult to reproduce the experiment. However, the proposed method converges to a certain result through repeated experiments, it is possible to reproduce and more accurate. Furthermore, the proposed method can predict genes that are highly related to disease among genes that are not predicted by the existing methods and discover novel relationships based on daily produced bibliographic data.