scAAGA: Single cell data analysis framework using asymmetric autoencoder with gene attention.

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Single-cell RNA sequencing (scRNA-seq) is crucial for understanding cellular heterogeneity.
• Analyzing scRNA-seq data, especially for COVID-19 research, presents significant challenges.
• Accurate cell clustering is essential for interpreting scRNA-seq datasets.

Purpose of the Study:

• To introduce scAAGA, a novel framework for enhanced scRNA-seq data analysis.
• To improve the accuracy and reliability of single-cell clustering using deep learning.
• To apply and validate scAAGA on COVID-19 peripheral blood mononuclear cell (PBMC) data.

Main Methods:

• Utilized an asymmetric autoencoder with a gene attention module for adaptive feature learning.
• Implemented data augmentation techniques to expand datasets and enhance model accuracy.
• Evaluated scAAGA performance against state-of-the-art methods using established metrics.

Main Results:

• scAAGA demonstrated superior performance in cell clustering compared to existing methods.
• Achieved significant improvements in Normalized Mutual Information (NMI) scores, ranging from 2.8% to 27.8%.
• Consistently outperformed other methods in Adjusted Rand Index (ARI) and Adjusted Mutual Information (AMI) scores.

Conclusions:

• scAAGA is a robust and effective tool for scRNA-seq data analysis.
• The framework enhances the accuracy and reliability of cell clustering, particularly in the context of COVID-19 research.
• Adaptive gene feature learning and data augmentation contribute to scAAGA's improved performance.