Inferring Novel Tumor Suppressor Genes with a Protein-Protein Interaction Network and Network Diffusion Algorithms

Area of Science:

• Computational Biology
• Genomics
• Cancer Research

Background:

• Identifying tumor suppressor genes (TSGs) is crucial for understanding cancer pathogenesis and developing treatments.
• Traditional experimental methods for TSG identification are time-consuming and difficult.
• Computational approaches offer a promising alternative for efficient TSG discovery.

Purpose of the Study:

• To develop and evaluate novel computational methods for identifying potential tumor suppressor genes.
• To compare the efficacy of two network diffusion algorithms: Laplacian heat diffusion (LHD) and random walk with restart (RWR).

Main Methods:

• Proposed two computational methods integrating LHD and RWR algorithms to search for candidate genes within a biological network.
• Implemented three strict screening tests to filter and increase the reliability of the identified putative TSGs.
• Compared the gene sets identified by the LHD-based and RWR-based methods.

Main Results:

• Identified a total of 169 potential TSGs, including 12 common genes (e.g., MAP3K10, RND1, OTX2).
• The LHD-based method uniquely identified 29 genes (e.g., RFC2, GUCY2F).
• The RWR-based method uniquely identified 128 genes (e.g., SNAI2, FGF4), with distinct findings from previous studies.

Conclusions:

• The proposed LHD-based and RWR-based computational methods are effective for identifying potential tumor suppressor genes.
• Some identified genes have been validated as novel TSGs in recent publications, confirming the utility of the methods.
• The identified gene sets differ from those reported previously, highlighting the novelty and potential of this approach.