MTGDC: A Multi-Scale Tensor Graph Diffusion Clustering for Single-Cell RNA Sequencing Data.

Area of Science:

• Computational Biology
• Genomics
• Bioinformatics

Background:

• Single-cell RNA sequencing (scRNA-seq) technology reveals cellular heterogeneity by analyzing gene expression in individual cells.
• Accurate computational methods are crucial for identifying distinct cell types within complex scRNA-seq datasets.

Purpose of the Study:

• To develop an advanced computational method for improved cell type detection in scRNA-seq data.
• Introduce the Multi-scale Tensor Graph Diffusion Clustering (MTGDC) algorithm for enhanced single-cell data analysis.

Main Methods:

• A multi-scale affinity learning method constructs a fully connected cell graph to capture similarity distributions.
• An efficient tensor graph diffusion learning framework processes multi-scale affinity matrices to learn high-order relationships.
• A fusion high-order affinity matrix is generated by combining multi-scale tensor graphs for spectral clustering.

Main Results:

• MTGDC effectively mines potential similarity distributions among cells.
• The tensor graph diffusion framework preserves both local high-order and global topology structure information.
• Experimental results demonstrate MTGDC's superior performance over existing algorithms.

Conclusions:

• MTGDC offers significant advantages in robustness, accuracy, visualization, and speed for scRNA-seq data analysis.
• The proposed method advances the field of computational single-cell biology by providing a more powerful clustering tool.