CDI

CDI evaluates clustering of cells by modeling scRNA-seq UMI counts as an unsupervised metric for assessing clustering performance.

Key Features:

• Unsupervised metric: CDI operates without ground truth labels to evaluate clustering results.
• UMI count modeling: CDI is based on explicit modeling of scRNA-seq UMI counts rather than transformed or normalized data.

Scientific Applications:

• scRNA-seq clustering evaluation: CDI assesses the accuracy of cell clustering into subpopulations in single-cell RNA-sequencing analyses.

Methodology:

CDI models scRNA-seq UMI counts to compute a clustering deviation index that quantifies clustering quality.