MOGAN for LUAD Subtype Classification by Integrating Three Omics Data Types.

Area of Science:

• Oncology
• Bioinformatics
• Genomics

Background:

• Lung adenocarcinoma (LUAD) is a heterogeneous cancer with poor prognosis, necessitating accurate subtype identification for effective treatment.
• Traditional single-omics approaches inadequately capture LUAD's molecular complexity.
• Multi-omics integration offers a comprehensive strategy to overcome single-omics limitations.

Purpose of the Study:

• To develop an advanced computational method for integrating multi-omics data in LUAD.
• To identify novel LUAD immune subtypes based on integrated molecular profiles.
• To assess the clinical relevance of identified subtypes for treatment guidance.

Main Methods:

• Utilized a Generative Adversarial Network (GAN), specifically the MOGAN method, to integrate transcriptomic, proteomic, and epigenomic data.
• Incorporated gene-protein and methylation-gene interactions within the MOGAN framework for enhanced data complementarity.
• Applied immune cell infiltration analysis to the integrated dataset for subtype discovery.

Main Results:

• Identified two distinct LUAD immune subtypes: MOGANTPM_S1 and MOGANTPM_S2.
• MOGANTPM_S1 exhibited higher immune infiltration, better prognosis, and sensitivity to immune checkpoint inhibitors (ICIs).
• MOGANTPM_S2 showed lower immune infiltration, poorer prognosis, and insensitivity to ICIs, suggesting immunotherapy is more suitable for MOGANTPM_S1.
• Developed a diagnostic model for LUAD subtypes using five key genes' transcriptomic and proteomic features.

Conclusions:

• The MOGAN method successfully integrated multi-omics data to identify LUAD immune subtypes with significant prognostic differences.
• This novel classification approach holds potential for guiding clinical treatment decisions in LUAD.
• The identified subtypes offer a basis for personalized immunotherapy strategies in lung adenocarcinoma.