gCAnno

gCAnno performs cell type annotation for single-cell transcriptome sequencing by constructing a cell type-gene bipartite graph, applying graph embedding to identify cell type-specific genes, and classifying single cells with naïve Bayes and Support Vector Machine classifiers.

Key Features:

• Graph-Based Methodology: Constructs a cell type-gene bipartite graph and uses graph embedding to extract cell type-specific genes.
• Machine Learning Classifiers: Implements two classifiers—gCAnno-Bayes (naïve Bayes) and gCAnno-SVM (Support Vector Machine)—for cell-level annotation.
• Cell Type-Specific Gene Identification: Leverages embedding-derived features to identify genes specific to each cell type for downstream classification.
• Benchmarking: Comparative studies reported higher accuracy and robustness than other state-of-the-art methods across multiple single-cell datasets with varying noise levels and platform differences.

Scientific Applications:

• Single-cell RNA analysis: Enables precise cell type identification in single-cell RNA/transcriptome sequencing to support studies of cellular heterogeneity and function.
• Cross-platform and noisy datasets: Applicable to datasets generated from different platforms and with varying noise levels to improve annotation robustness.

Methodology:

Graph construction of a cell type-gene bipartite graph; graph embedding to extract cell type-specific genes; classification using gCAnno-Bayes (naïve Bayes) and gCAnno-SVM (Support Vector Machine) based on the embedded gene data.