scDBic: a novel deep learning-based biclustering algorithm for analyzing scRNA-seq data.

Area of Science:

• Computational Biology
• Genomics
• Bioinformatics

Background:

• Clustering single-cell RNA sequencing (scRNA-seq) data is crucial for understanding cellular heterogeneity.
• Existing clustering algorithms struggle with local consistency, while biclustering methods face challenges like cell loss and adaptability to high-dimensional data.

Purpose of the Study:

• To introduce scDBic, a novel deep learning-based biclustering algorithm designed for scRNA-seq data.
• To improve cell clustering performance by effectively capturing gene expression information and identifying key genes within cell groups.

Main Methods:

• scDBic employs a three-step process: cell clustering using a deep autoencoder, gene clustering, and identification of key gene clusters via a reverse strategy.
• The deep autoencoder captures essential gene expression patterns, while the reverse strategy pinpoints genes specific to identified cell groups.

Main Results:

• The algorithm successfully discovers distinct cell groups within scRNA-seq datasets.
• scDBic identifies the key genes associated with each discovered cell group.
• Performance evaluation shows scDBic outperforms traditional clustering and biclustering algorithms in scRNA-seq data analysis.

Conclusions:

• scDBic offers a powerful new approach for analyzing scRNA-seq data, enabling simultaneous cell group discovery and key gene identification.
• This technique provides a direct and effective method for exploring cellular heterogeneity and gene functions in complex biological systems.