scClassifR

scClassifR classifies cell types in single-cell RNA-sequencing (scRNA-seq) datasets using hierarchically organized support vector machines (SVMs) within an R-based framework to provide robust, probability-based and extensible cell-type identification.

Key Features:

• Automatic Cell Type Classification: Automates identification of cell types from scRNA-seq gene expression profiles using trained classifiers.
• R-based Compatibility: Accepts Seurat and Bioconductor SingleCellExperiment objects as input for integration into R analysis workflows.
• Hierarchical SVM Approach: Uses hierarchically organized Support Vector Machines (SVMs) to distinguish specific cell types against all other cells.
• Robustness to Unknown Cell Types: Implements behavior to avoid misclassifying cells not represented in the reference, reducing false positives.
• Ambiguous Assignment Reporting: Reports ambiguous cell assignments together with associated probabilities to convey classification confidence.
• Extensibility and Customization: Provides functions for training and evaluating classifiers to add or refine models for additional cell types.

Scientific Applications:

• Developmental Biology: Enables identification and tracking of cell types and differentiation states in developing tissues using scRNA-seq data.
• Cancer Research: Supports classification of tumor and microenvironment cell populations to dissect cellular composition in cancer samples.
• Immunology: Facilitates annotation of immune cell subsets and states from single-cell transcriptomic datasets.
• Cellular Heterogeneity and Discovery: Assists in resolving cellular heterogeneity and identifying novel cell types or states within complex tissues.

Methodology:

Hierarchically organized Support Vector Machines (SVMs) classify cells based on gene expression profiles; the package reports ambiguous assignments with probabilities and provides functions to train and evaluate classifiers on additional cell types, accepting Seurat and SingleCellExperiment objects as input.