SAME-clustering: Single-cell Aggregated Clustering via Mixture Model Ensemble.

Area of Science:

• Computational Biology
• Genomics
• Bioinformatics

Background:

• Clustering is crucial for analyzing single-cell RNA sequencing (scRNA-seq) data to identify cell types and their transcriptomic signatures.
• Existing clustering methods provide variable results for cell type and cluster assignments, lacking a universally superior approach.
• The challenge lies in selecting the optimal method for diverse scRNA-seq datasets.

Purpose of the Study:

• To develop an improved ensemble clustering method for scRNA-seq data analysis.
• To address the variability and challenges in unsupervised clustering of single-cell data.
• To enhance the accuracy of cell type identification and cluster assignments.

Main Methods:

• Developed SAME-clustering, a mixture model-based approach for ensemble clustering.
• Input: Clustering solutions from multiple existing methods.
• Output: A maximally diverse subset of solutions to generate an improved ensemble clustering result.

Main Results:

• SAME-clustering was evaluated on 15 diverse scRNA-seq datasets.
• Demonstrated enhanced clustering performance in both cluster assignments and determination of the number of clusters.
• The method showed consistent improvements across datasets with varying cell numbers and cluster counts.

Conclusions:

• SAME-clustering provides a robust and effective ensemble approach for scRNA-seq data.
• The mixture model ensemble clustering framework is adaptable to various clustering applications beyond scRNA-seq.
• This method offers a valuable tool for more accurate cell type deconvolution in complex biological samples.