SinCWIm: An imputation method for single-cell RNA sequence dropouts using weighted alternating least squares

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Single-cell RNA sequencing (scRNA-seq) is crucial for analyzing cellular heterogeneity and differentiation.
• Technological limitations in scRNA-seq lead to dropout events, introducing noise and affecting data accuracy.
• Accurate imputation of dropout events is essential for reliable downstream scRNA-seq analyses.

Purpose of the Study:

• To develop a robust method for identifying and imputing dropout events in scRNA-seq data.
• To improve the performance of clustering, cell annotation, and differential gene expression analysis by addressing data noise.

Main Methods:

• Proposed SinCWIm method utilizing weighted alternating least squares (WALS).
• Quantified zero-entry confidence using Pearson correlation and hierarchical clustering.
• Integrated WALS with outlier removal and data correction for accurate imputation.

Main Results:

• SinCWIm demonstrated superior performance across eight scRNA-seq datasets compared to existing methods.
• Achieved high adjusted RAND index scores for clustering on Usoskin (94.46%), Pollen (96.48%), and Bladder (76.74%) datasets.
• Showcased significant improvements in differential gene expression retention and data visualization.

Conclusions:

• SinCWIm offers a valuable imputation solution for scRNA-seq dropout events.
• The method exhibits excellent performance in clustering and visualization across diverse scRNA-seq datasets.
• SinCWIm is a versatile tool applicable to various single-cell sequencing data types.