GAiN: An integrative tool utilizing generative adversarial neural networks for augmented gene expression analysis.

Area of Science:

• Genomics
• Bioinformatics
• Artificial Intelligence in Medicine

Background:

• Precision medicine relies on analyzing large genomic datasets.
• Studying rare diseases and under-represented populations presents challenges due to small sample sizes.
• Existing biostatistical methods struggle with limited data for gene expression analysis.

Purpose of the Study:

• To develop an integrative approach for gene expression analysis in small datasets.
• To address the limitations of conventional methods in identifying differentially expressed genes (DEGs) in small cohorts.
• To leverage big population data to enhance the analysis of limited sample sizes.

Main Methods:

• Developed GAiN, an integrative approach using an ensemble of generative adversarial networks (GANs).
• GAiN captures gene expression patterns from small datasets by leveraging big population data.
• Benchmarked GAiN against a gold standard for identifying DEGs and enriched pathways.

Main Results:

• GAiN reliably discovers DEGs and enriched pathways between cohorts with as few as 10 samples.
• The method outperforms conventional biostatistical approaches when data is limited.
• Demonstrated the effectiveness of GAiN in scenarios with small sample sizes.

Conclusions:

• GAiN provides a reliable tool for gene expression analysis in limited-sample scenarios.
• The approach is crucial for studying rare diseases, under-represented populations, and resource-limited research.
• GAiN is freely available, facilitating its adoption in clinical research.