ML-GAP: machine learning-enhanced genomic analysis pipeline using autoencoders and data augmentation.

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• RNA sequencing (RNA-Seq) generates complex, high-volume data.
• Identifying differentially expressed genes (DEGs) is crucial for understanding diseases like cancer.
• Current methods face challenges in analyzing large-scale transcriptomic data.

Purpose of the Study:

• To introduce a novel Machine Learning-Enhanced Genomic Data Analysis Pipeline (ML-GAP).
• To improve the accuracy and efficiency of DEG identification from RNA-Seq data.
• To leverage advanced machine learning techniques for genomic data analysis.

Main Methods:

• Development of the ML-GAP pipeline incorporating autoencoders.
• Implementation of innovative data augmentation strategies, specifically the MixUp method.
• Utilizing MixUp to create synthetic training data via linear combinations for improved model generalization.

Main Results:

• ML-GAP demonstrated superior accuracy, efficiency, and insight generation compared to existing methods.
• The MixUp method significantly contributed to the pipeline's enhanced performance.
• The study highlights advancements in genomic data analysis and DEG detection.

Conclusions:

• ML-GAP enables more accurate detection of DEGs, advancing genomic data analysis.
• The pipeline offers potential for new therapeutic interventions and research avenues.
• Integration of explainable AI (XAI) ensures transparent and interpretable genetic marker identification.