GiniClust2

GiniClust2 identifies common and rare cell types in single-cell gene expression data by integrating the Gini index and Fano factor into a cluster-aware weighted ensemble clustering framework.

Key Features:

• Data type: Operates on single-cell gene expression data.
• Gini index: Assesses inequality in gene expression distributions across cells.
• Fano factor: Captures gene expression variability and aids identification of rare events.
• Cluster-aware weighted ensemble clustering: Integrates Gini and Fano metrics into a weighted ensemble that forms clusters with awareness of their composition.
• Detection sensitivity: Simultaneously identifies prevalent (common) and infrequent (rare) cellular signatures.
• Validation and scalability: Validated on diverse datasets and demonstrates scalability for large-scale single-cell gene expression data.

Scientific Applications:

• Developmental biology: Identifying diverse cell populations during tissue or organ development.
• Oncology: Detecting rare tumor subpopulations alongside common cancer cell types.
• Immunology: Profiling immune cell heterogeneity and rare immune subsets.
• Tissue and organ heterogeneity: Mapping cellular composition within tissues or organs to study complex biological processes and disease mechanisms.

Methodology:

Computes Gini index and Fano factor and integrates these metrics using a cluster-aware weighted ensemble clustering approach that forms clusters with awareness of their composition to detect common and rare cell types.