Mcadet: A feature selection method for fine-resolution single-cell RNA-seq data based on multiple correspondence...

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Single-cell RNA sequencing (scRNA-seq) data analysis is complex due to high dimensionality, sparsity, and noise.
• Existing methods for selecting highly variable genes (HVGs) show inconsistencies and struggle with fine-resolution or rare cell types.
• Reliable HVG selection is crucial for accurate downstream scRNA-seq data analysis.

Purpose of the Study:

• To introduce Mcadet, a novel feature selection framework for scRNA-seq data.
• To address limitations of current HVG selection methods, particularly for challenging datasets.
• To improve the reliability and accuracy of identifying informative genes in scRNA-seq analysis.

Main Methods:

• Mcadet integrates Multiple Correspondence Analysis (MCA), graph-based community detection, and a novel statistical testing approach.
• The framework was evaluated using both simulated and real-world scRNA-seq datasets.
• Unbiased metrics were employed to compare Mcadet's performance against existing methods.

Main Results:

• Mcadet demonstrated superior performance in selecting HVGs from fine-resolution scRNA-seq datasets.
• The framework accurately identified HVGs in datasets containing minority cell populations.
• Mcadet enhances the reliability of selected HVGs compared to existing approaches.

Conclusions:

• Mcadet offers a robust solution for HVG selection in scRNA-seq data, particularly for complex biological scenarios.
• The improved HVG selection by Mcadet can enhance the reliability of downstream analyses.
• Further investigation is needed to fully understand the impact of Mcadet's HVG selection on diverse downstream applications.