Single-Cell Transcriptome Data Clustering via Multinomial Modeling and Adaptive Fuzzy K-Means Algorithm.

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Single-cell RNA sequencing (scRNA-seq) provides cellular resolution for studying tissue heterogeneity.
• Droplet-based sequencing enables high-throughput processing of thousands of single cells.
• Clustering sparse, high-dimensional scRNA-seq UMI count data is challenging due to noise and dimensionality.

Purpose of the Study:

• To develop a novel and effective clustering method for single-cell transcriptome UMI count data.
• To address limitations of existing deep learning methods in modeling gene expression profiles and data structure.
• To improve the accuracy and scalability of clustering large-scale single-cell datasets.

Main Methods:

• Combined deep autoencoder technique with statistical modeling for scDMFK.
• Utilized multinomial distribution to characterize data structure and neural networks for parameter estimation.
• Employed an adaptive fuzzy k-means algorithm with entropy regularization for soft clustering in a latent space.

Main Results:

• scDMFK accurately models single-cell UMI count data structure using multinomial distribution.
• The method outperforms existing state-of-the-art approaches in data modeling and clustering accuracy across simulations and real datasets.
• scDMFK demonstrates excellent scalability for analyzing large-scale single-cell datasets.

Conclusions:

• scDMFK offers a robust and scalable solution for clustering single-cell RNA sequencing data.
• The integration of deep learning with statistical modeling provides superior performance in capturing gene expression profiles.
• This method enhances the analysis of cellular heterogeneity and cell type identification in complex biological systems.