G2S3: A gene graph-based imputation method for single-cell RNA sequencing data.

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Single-cell RNA sequencing (scRNA-seq) offers high-resolution gene expression analysis.
• Data sparsity and dropout events are significant challenges in scRNA-seq data, hindering downstream analysis.
• Existing imputation methods often struggle to accurately recover true gene expression levels.

Purpose of the Study:

• To develop and validate a novel imputation method, G2S3, for addressing dropout events in scRNA-seq data.
• To evaluate the performance of G2S3 against existing imputation techniques across diverse scRNA-seq datasets.
• To demonstrate the utility of G2S3 in improving key scRNA-seq analysis tasks.

Main Methods:

• G2S3 imputes dropouts by leveraging information from adjacent genes within a sparse gene graph constructed from scRNA-seq profiles.
• The method was applied to eight distinct scRNA-seq datasets.
• Performance was benchmarked against ten established imputation methods.

Main Results:

• G2S3 demonstrated superior performance in recovering gene expression compared to existing methods.
• The method significantly improved the identification of cell subtypes and the reconstruction of cell trajectories.
• G2S3 effectively enhanced the discovery of differentially expressed genes and the recovery of gene regulatory networks.
• Computational efficiency was observed, making G2S3 suitable for large-scale datasets.

Conclusions:

• G2S3 is a highly effective and computationally efficient method for imputing dropout events in scRNA-seq data.
• The proposed method substantially improves the accuracy and reliability of various downstream analyses in single-cell transcriptomics.
• G2S3 offers a robust solution for overcoming data sparsity challenges, advancing the utility of scRNA-seq technology.