Neural Latent Filtering for Gene Discovery in Breast Cancer Subtypes.

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• High-dimensional gene expression data poses challenges for identifying cancer-relevant genes.
• Filtering methods are crucial for selecting informative genes as potential biomarkers.
• Existing methods may struggle with the complexity of genomic data for cancer subtype analysis.

Purpose of the Study:

• To propose a novel neural-based filtering approach for gene selection from RNA sequencing data.
• To identify key genes associated with specific breast cancer subtypes (Luminal-A and Basal-like).
• To enhance the understanding of the molecular landscape of different breast cancer subtypes.

Main Methods:

• Utilized a neural network-based filtering approach to extract latent gene representations.
• Applied the method to a breast invasive carcinoma dataset.
• Focused on identifying differentially relevant genes between Luminal-A and Basal-like breast cancer subtypes.

Main Results:

• The neural-based approach successfully identified a subset of informative genes.
• The selected genes provide insights into the distinct molecular characteristics of Luminal-A and Basal-like breast cancer.
• Demonstrated the potential of latent representations for effective gene filtering.

Conclusions:

• The proposed neural-based filtering method is effective for gene selection in high-dimensional expression data.
• This approach can aid in identifying novel biomarkers for breast cancer subtypes.
• Further investigation into the identified genes can advance personalized cancer therapy strategies.