Predicting non-small cell lung cancer-related genes by a new network-based machine learning method

Area of Science:

• Genomics
• Computational Biology
• Oncology

Background:

• Lung cancer is a leading global cause of cancer death, with non-small cell lung cancer (NSCLC) accounting for over 85% of cases.
• Family history and genetic studies (NGS, GWAS) have identified some NSCLC-related genes, but often overlook complex gene interactions and can yield false positives.
• Existing methods for disease-gene association are costly and may miss crucial interaction data.

Purpose of the Study:

• To develop a novel, cost-effective computational method for identifying genes associated with NSCLC.
• To address limitations of existing genetic association studies, such as high costs and false-positive rates.
• To explore gene-gene interactions in the context of NSCLC using network-based machine learning.

Main Methods:

• Proposed a network-based machine learning method named deepRW for NSCLC gene prediction.
• Constructed a gene interaction network incorporating NSCLC-related and unrelated genes.
• Employed deep walk and graph convolutional network (GCN) for learning gene-disease interactions, followed by a deep neural network (DNN) for gene prediction.

Main Results:

• The deepRW method demonstrated superior performance compared to existing approaches in predicting NSCLC-related genes.
• 10-fold cross-validation confirmed the high accuracy and effectiveness of the proposed deepRW method.
• Comparative experiments validated the contribution of each module within the deepRW framework.

Conclusions:

• The deepRW method provides a powerful and efficient tool for discovering novel NSCLC-associated genes.
• This network-based machine learning approach offers a promising alternative for low-cost, high-accuracy disease-gene association studies.
• The findings highlight the importance of considering gene interactions in NSCLC research.