gCAnno

gCAnno introduces a novel graph-based approach to the challenge of cell type annotation in single-cell RNA analysis. It aims to enhance accuracy and overcome the limitations of current methods, which primarily rely on cluster-level annotations. This method addresses the variability and potential inaccuracies introduced by multiple clustering algorithms and the requirement for numerous parameter adjustments, often resulting in incorrect or inconsistent cluster-level annotations and necessitating multiple clustering iterations.

By constructing a cell type-gene bipartite graph and applying graph embedding techniques, gCAnno efficiently identifies cell type-specific genes. It then utilizes these genes to build classification models using both naive Bayes (gCAnno-Bayes) and Support Vector Machine (SVM) (gCAnno-SVM) approaches, facilitating precise single-cell level annotations.