CALLR: a semi-supervised cell-type annotation method for single-cell RNA sequencing data.

Area of Science:

• Computational biology
• Genomics
• Bioinformatics

Background:

• Single-cell RNA sequencing (scRNA-seq) enables comprehensive analysis of cellular heterogeneity.
• Accurate cell-type annotation is crucial for interpreting scRNA-seq data.
• Traditional methods relying on clustering and marker genes often lack precision.

Purpose of the Study:

• To develop a more accurate and robust cell-type annotation method for scRNA-seq data.
• To introduce CALLR, a semi-supervised learning approach for improved annotation.
• To validate CALLR's performance against existing annotation techniques.

Main Methods:

• CALLR integrates unsupervised learning (graph Laplacian matrix) with supervised learning (sparse logistic regression).
• The method employs an iterative optimization process to refine cell clusters and annotation labels.
• A computationally efficient algorithm is developed for solving the optimization problem.

Main Results:

• CALLR demonstrates high accuracy in cell-type annotation across 10 real scRNA-seq datasets.
• The proposed method outperforms traditional clustering-based approaches.
• CALLR shows superior performance compared to other (semi-)supervised learning methods.

Conclusions:

• CALLR offers a significant advancement in scRNA-seq data analysis for cell-type annotation.
• The semi-supervised approach effectively leverages both unsupervised and supervised learning principles.
• CALLR provides a reliable and accurate tool for researchers studying cellular heterogeneity.