scAnnotatR

scAnnotatR classifies cell types in single-cell RNA sequencing (scRNA-seq) datasets using pre-trained, hierarchical Support Vector Machine classifiers to provide accurate and scalable automated cell-type annotation.

Key Features:

• Compatibility with R data structures: Integrates with Seurat and Bioconductor's SingleCellExperiment data structures.
• Hierarchical SVM Framework: Employs a hierarchical organization of Support Vector Machines (SVMs) to distinguish specific cell types from all others.
• Pre-trained classifiers: Uses pre-trained classifiers for annotation to enhance accuracy and efficiency.
• Handling large datasets: Scales to process datasets containing over 600,000 cells.
• Robust classification performance: Achieves high accuracy, sensitivity, and specificity in cell-type classification.
• Unknown-cell handling: Can leave cells unclassified when reference datasets lack corresponding cell-type representations.

Scientific Applications:

• Developmental biology: Enables precise cell-type identification in developmental biology studies using scRNA-seq data.
• Immunology: Supports immune-cell profiling and annotation in immunology research.
• Cancer genomics: Facilitates tumor and microenvironment cell-type annotation in cancer genomics studies.

Methodology:

Uses pre-trained classifiers within a hierarchical Support Vector Machine framework to distinguish specific cell types from others and to flag cells absent from reference datasets as unknown.