Imputing dropouts for single-cell RNA sequencing based on multi-objective optimization

Area of Science:

• Genomics and Bioinformatics
• Computational Biology
• Single-cell analysis

Background:

• Single-cell RNA sequencing (scRNA-seq) provides high-resolution transcriptome profiling.
• Technical noise in scRNA-seq leads to excess zeros (dropouts), compromising downstream analysis accuracy.
• Accurate imputation methods are essential for correcting scRNA-seq data dropouts.

Purpose of the Study:

• To develop a novel dropout imputation method for scRNA-seq data.
• To address limitations of existing methods that rely on preconceived data structures.
• To improve the accuracy of downstream analyses by effectively handling scRNA-seq data dropouts.

Main Methods:

• Developed a multi-objective optimization framework for dropout imputation.
• Assumed data integrates horizontal, vertical, and low-rank latent structures.
• Learned latent structures and combination weights via multi-objective optimization.

Main Results:

• The proposed method demonstrates superior performance in recovering true gene expression profiles.
• Significant improvements observed in differential expression analysis, cell clustering, and cell trajectory inference.
• The method effectively addresses dropout issues in scRNA-seq data.

Conclusions:

• The multi-objective optimization approach offers a robust solution for scRNA-seq data imputation.
• This method enhances the reliability and accuracy of various single-cell data analyses.
• The developed R package (scMOO) and associated codes are publicly available.