scEVE: a single-cell RNA-seq ensemble clustering algorithm capitalizing on the differences of predictions between...

Area of Science:

• Computational biology
• Genomics
• Bioinformatics

Background:

• Single-cell RNA sequencing (scRNA-seq) is crucial for analyzing individual cell transcriptomes.
• Numerous clustering methods exist for scRNA-seq data, but their predictions vary due to differing hypotheses.
• Current ensemble algorithms integrate multiple methods but do not provide uncertainty or multiple resolutions.

Purpose of the Study:

• To introduce a novel ensemble clustering approach for scRNA-seq data.
• To address limitations of existing methods by generating clustering results with uncertainty values and multiple resolutions.
• To present the scEVE algorithm as an advancement in single-cell data analysis.

Main Methods:

• Developed the scEVE algorithm, an original ensemble clustering approach.
• Focused on describing differences between clustering results rather than minimizing them.
• Evaluated scEVE on 15 experimental and 1200 synthetic scRNA-seq datasets.

Main Results:

• scEVE outperforms the current state-of-the-art ensemble clustering methods.
• The algorithm successfully generates clustering results with uncertainty values and multiple resolutions.
• Demonstrated the benefits of addressing these conceptual challenges for biological downstream analyses.

Conclusions:

• scEVE offers a significant improvement over existing single-cell ensemble clustering techniques.
• The approach of describing differences between clustering results provides valuable insights.
• This work sets a new direction for developing advanced single-cell ensemble clustering algorithms.