DCRELM: dual correlation reduction network-based extreme learning machine for single-cell RNA-seq data clustering.

Area of Science:

• Genomics
• Computational Biology
• Bioinformatics

Background:

• Single-cell ribonucleic acid sequencing (scRNA-seq) enables detailed transcriptome analysis.
• Cluster analysis is crucial for understanding cell heterogeneity in scRNA-seq data.
• Existing clustering methods face challenges with high dimensionality, noise, and sparsity in scRNA-seq data.

Purpose of the Study:

• To develop an advanced clustering algorithm for scRNA-seq data.
• To overcome the limitations of current clustering approaches in handling complex scRNA-seq datasets.
• To enhance the accuracy and robustness of cell clustering.

Main Methods:

• Proposed Dual Correlation Reduction network-based Extreme Learning Machine (DCRELM) algorithm.
• Utilized extreme learning machine (ELM) for low-dimensional feature extraction.
• Employed ELM graph distortion for feature robustness and autoencoder fusion for latent representation.
• Incorporated dual information reduction and triplet self-supervised learning.

Main Results:

• DCRELM demonstrated superior clustering performance compared to existing methods.
• The algorithm showed enhanced robustness in analyzing scRNA-seq data.
• Extensive experiments validated the effectiveness of DCRELM.

Conclusions:

• DCRELM offers a powerful solution for clustering scRNA-seq data.
• The proposed method effectively addresses challenges of dimensionality, noise, and sparsity.
• DCRELM provides a robust and accurate tool for single-cell data analysis.