OnClass

OnClass classifies single-cell transcriptomic data into Cell Ontology terms and identifies marker genes to harmonize and annotate cell types across datasets.

Key Features:

• Automated Classification: Automatically classifies single-cell datasets into Cell Ontology terms to harmonize annotations across datasets.
• Inference Beyond Training Data: Infers cell types not present in the training set by exploiting the hierarchical structure of the Cell Ontology graph.
• Marker Gene Identification: Identifies marker genes for all Cell Ontology categories regardless of representation in the training data.
• Refinement of the Cell Ontology: Uses marker genes and classification results to support data-driven refinement and potential expansion of the Cell Ontology.

Scientific Applications:

• Single-cell genomics and transcriptomics: Annotating cell types in single-cell RNA-seq and other single-cell transcriptomic datasets.
• Meta-analysis and integration: Harmonizing annotations across datasets to enable meta-analyses and large-scale integrative studies.
• Cellular heterogeneity and discovery: Characterizing cellular heterogeneity and facilitating discovery of novel or rare cell types.
• Developmental biology and disease studies: Investigating developmental processes and disease mechanisms through consistent cell-type annotation and marker discovery.

Methodology:

Maps single-cell data onto the Cell Ontology graph and integrates machine learning with ontological frameworks, using supervised learning for classification and unsupervised methods to infer relationships between cell types.