Bubble: a fast single-cell RNA-seq imputation using an autoencoder constrained by bulk RNA-seq data.

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Single-cell RNA sequencing (scRNA-seq) provides high-resolution gene expression data.
• scRNA-seq data suffers from sparsity, characterized by 'dropout' events (unobserved gene expression).
• Dropout events negatively impact downstream analyses like clustering and differential expression.

Purpose of the Study:

• To develop a method for identifying and imputing dropout events in scRNA-seq data.
• To improve the accuracy and reliability of scRNA-seq data analysis.
• To enhance downstream scRNA-seq applications.

Main Methods:

• Proposed Bubble, a novel method for dropout event identification and imputation.
• Bubble identifies dropouts using gene expression rate and coefficient of variation within cell subpopulations.
• Employs an autoencoder constrained by bulk RNA sequencing data for imputation, reducing false positives.

Main Results:

• Bubble accurately identifies and imputes missing values in scRNA-seq data.
• Demonstrated enhanced gene-to-gene and cell-to-cell correlations.
• Reduced the introduction of false positive signals compared to other methods.
• Improved performance in differential expression analysis, clustering, and trajectory inference.

Conclusions:

• Bubble offers a fast, scalable, and memory-efficient solution for scRNA-seq data imputation.
• The method effectively mitigates the impact of dropout events on scRNA-seq analysis.
• Bubble enhances the utility of scRNA-seq data for biological discovery.