Integrating Deep Supervised, Self-Supervised and Unsupervised Learning for Single-Cell RNA-seq Clustering and...

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Single-cell RNA sequencing (scRNA-seq) technologies generate massive gene expression datasets.
• Cell clustering and annotation are critical but challenging downstream analyses.
• Existing methods often fail to leverage valuable cell type labels from reference datasets.

Purpose of the Study:

• To develop a novel framework, scAnCluster, for supervised cell clustering and annotation.
• To utilize existing cell type labels from reference data to guide analysis of unlabeled target data.
• To improve the accuracy and efficiency of scRNA-seq data analysis, including the discovery of novel cell types.

Main Methods:

• Integration of deep supervised learning, self-supervised learning, and unsupervised learning techniques.
• An end-to-end framework designed for supervised cell clustering and annotation.
• Leveraging cell type labels from reference datasets to enhance clustering and annotation of target datasets.

Main Results:

• scAnCluster outperforms existing supervised scRNA-seq clustering methods on both simulated and real-world data.
• The framework effectively utilizes reference cell type labels for improved clustering and annotation.
• Demonstrated success in identifying novel cell types not present in the reference data.

Conclusions:

• scAnCluster offers a powerful and versatile approach for scRNA-seq data analysis.
• The framework enhances the reliability of cell clustering and annotation by integrating external label information.
• scAnCluster facilitates the discovery of new biological insights from complex single-cell datasets.