ccImpute

ccImpute corrects dropout events in single-cell RNA sequencing (scRNA-seq) data by imputing zeros from lowly expressed genes to improve the accuracy of downstream gene expression analyses.

Key Features:

• Dropout Event Correction: Identifies and imputes zeros arising from dropout events in scRNA-seq to restore signals from lowly expressed genes.
• Consensus clustering-based similarity measure: Employs consensus clustering to derive cell-to-cell similarity and uses those similarities to predict and correct dropout values.
• Efficiency and performance: Demonstrates improved clustering characteristics on datasets with known cell identities and introduces minimal additional noise compared with existing imputation algorithms.
• Downstream analysis impact: Produces imputed expression matrices that facilitate analyses of cellular responses under conditions such as disease versus non-disease states and treatment protocols.

Scientific Applications:

• Cellular heterogeneity: Enables detection and characterization of subpopulations by recovering lowly expressed marker genes.
• Disease progression: Improves sensitivity to subtle expression changes relevant to disease staging and progression.
• Response to therapeutic interventions: Enhances detection of treatment-induced transcriptional changes at single-cell resolution.
• Comparative condition studies: Supports analyses comparing disease versus non-disease states and different treatment protocols by reducing dropout-related confounding.

Methodology:

Identifies and corrects dropout events by imputing zero counts; employs consensus clustering to measure cell similarity for imputation; validates performance by comparing clustering characteristics on datasets with known cell identities; implemented in R.