scHFC: a hybrid fuzzy clustering method for single-cell RNA-seq data optimized by natural computation

Area of Science:

• Computational Biology
• Bioinformatics
• Genomics

Background:

• Single-cell RNA sequencing (scRNA-seq) enables cellular-level biological exploration.
• Clustering is vital for understanding cell heterogeneity in scRNA-seq data.
• Existing methods struggle with scRNA-seq data due to dropout events and high dimensionality, impacting cell type accuracy.

Purpose of the Study:

• To introduce scHFC, a novel hybrid fuzzy clustering method for scRNA-seq data analysis.
• To enhance the accuracy and stability of cell clustering in scRNA-seq datasets.
• To provide a robust tool for downstream analysis of scRNA-seq data.

Main Methods:

• scHFC combines Fuzzy C-Means (FCM) and Gath-Geva (GG) algorithms, optimized by natural computation.
• Principal Component Analysis (PCA) is used for dimensionality reduction after data preprocessing.
• FCM is optimized using simulated annealing and genetic algorithms, with results feeding into the GG algorithm.
• A multi-index comprehensive estimation method is employed for accurate cluster number determination.

Main Results:

• The scHFC method demonstrated superior performance across 17 scRNA-seq datasets compared to six state-of-the-art methods.
• Experimental results validated improved clustering accuracy and algorithmic stability of scHFC.
• The method effectively addresses challenges posed by dropout events and high dimensionality in scRNA-seq data.

Conclusions:

• scHFC is an effective and robust method for clustering cells in scRNA-seq data.
• The proposed method shows significant potential for improving downstream analyses in single-cell genomics.
• The developed scHFC algorithm offers enhanced accuracy and stability for cell type identification.