Integrating Multiple Single-Cell RNA Sequencing Datasets Using Adversarial Autoencoders.

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Single-cell RNA sequencing (scRNA-seq) enables gene expression heterogeneity quantification.
• Integrating multiple scRNA-seq datasets requires batch effect correction.
• Current unsupervised methods do not leverage cell-type labels for improved batch correction.

Purpose of the Study:

• To propose a novel deep learning model, IMAAE, for batch effect correction in scRNA-seq data.
• To enhance batch correction performance by utilizing known single-cell cluster labeling information.
• To provide a robust method for integrating complex, multi-dataset scRNA-seq scenarios.

Main Methods:

• Developed IMAAE (integrating multiple single-cell datasets via an adversarial autoencoder), a deep learning model.
• Employed adversarial autoencoder architecture to learn data representations.
• Incorporated single-cell cluster labels into the model for supervised batch correction.

Main Results:

• IMAAE demonstrated superior performance over existing methods in qualitative and quantitative evaluations.
• The model effectively corrected batch effects across various dataset scenarios.
• IMAAE preserved both corrected dimension reduction data and gene expression data.

Conclusions:

• IMAAE offers an effective deep learning approach for batch effect correction in scRNA-seq data.
• Utilizing cell labels significantly improves batch correction, especially for complex datasets.
• IMAAE presents a promising new option for large-scale scRNA-seq data integration and analysis.