Missing data in single-cell transcriptomes reveals transcriptional shifts.

Area of Science:

• Single-cell genomics
• Computational biology
• Developmental biology

Background:

• Single-cell RNA sequencing (scRNA-seq) enables large-scale transcriptome profiling.
• Cell prioritization for downstream analysis is challenging due to technical variations and complex data structures.
• Existing methods are often confounded by clustering, normalization, and dimensionality reduction techniques.

Purpose of the Study:

• To develop a novel scoring approach for unbiased cell prioritization.
• To identify distinct transcriptomes based on missing data patterns, independent of common scRNA-seq processing steps.
• To enable robust feature selection for subsequent analyses.

Main Methods:

• A scoring approach was developed utilizing missing data patterns in single-cell transcriptomes.
• The method was designed to be independent of clustering, normalization, and dimensionality reduction strategies.
• The approach was applied to analyze transcriptomic data from Dictyostelium discoideum.

Main Results:

• The scoring method successfully identified distinct transcriptomes within a dataset.
• Unbiased cell prioritization and feature selection were achieved.
• Analysis of Dictyostelium discoideum revealed a significant metabolic shift associated with developmental state transitions.

Conclusions:

• The developed scoring approach offers a robust and unbiased method for cell prioritization in single-cell transcriptomics.
• This technique overcomes limitations of traditional methods confounded by data processing steps.
• The findings in Dictyostelium discoideum highlight the utility of the method for uncovering biological insights, such as developmental metabolic changes.