As one of the branches of natural science, the research of biology has never been stopped along with human history. Biology helps us understand the living world, and it is essential for many fields such as agriculture, animal husbandry, medicine, and pharmacy. Traditional biology is based on macroscopic experiments and statistics according to fundamental units of life. However, traditional biological experiments tend to be costly and time-consuming. As a major interdisciplinary area between informatics and life science, bioinformatics has achieved remarkable results in various fields. Nowadays, since there already exists a huge amount of bio-data, we can use bioinformatics methods to analyze or predict existing or unknown data to guide future experiments, which can save precious research resources.

In this presentation, two kinds of graph theory-based methods for analyzing biological networks will be introduced. The first study is the application of the minimum feedback vertex set (MFVS) in predicting human cancer genes. The proposed methods integrate the advantage of the traditional gene differential expression value and the MFVS method. The second study is about the prediction of the hot spot residues in protein complexes. In this study, several densest subgraph-based methods are considered to analyze residues interaction networks, and these methods have the advantage of recall and F-score in finding potential hot spot residues. In the end, a simple introduction about my future research plan will be discussed.