A semi-automatic cell type annotation method for single-cell RNA sequencing dataset.

Area of Science:

• Genomics
• Bioinformatics
• Cell Biology

Background:

• Single-cell RNA sequencing (scRNA-seq) enables detailed cell-by-cell expression analysis.
• Current cell type annotation methods are often time-consuming and subjective.
• Efficient analysis methods are critical due to the high dimensionality of scRNA-seq data.

Purpose of the Study:

• To develop a semi-automatic, efficient, and accurate method for cell type annotation in scRNA-seq data.
• To streamline the annotation process from laborious manual assessment to a more objective, score-based system.
• To validate the proposed method on a real-world biological dataset.

Main Methods:

• A novel semi-automatic method was developed to calculate normalized scores for cell types.
• The method utilizes user-defined, cell type-specific marker gene lists.
• Applied to mouse cardiac non-myocyte scRNA-seq data, involving t-stochastic neighbor embedding (t-SNE) clustering.

Main Results:

• Successfully annotated 35 t-SNE clusters into 12 distinct cell types.
• Annotation accuracy was confirmed through co-expression network construction for each cell type.
• Gene Ontology analysis and regulatory network analysis supported the annotated cell types and identified relevant transcription factors.

Conclusions:

• The proposed semi-automatic method offers a straightforward and accurate approach to cell type annotation for scRNA-seq data.
• The method's efficiency and objectivity address limitations of conventional annotation techniques.
• Validated findings with biological evidence, demonstrating the utility of the developed R script.