SDImpute

SDImpute performs dropout imputation for single-cell RNA sequencing (scRNA-seq) data to recover missing gene expression values and preserve cellular heterogeneity for downstream analyses such as clustering, visualization, and differential expression analysis.

Key Features:

• Block Imputation Methodology: Uses a block imputation strategy tailored to scRNA-seq that leverages both cell-level and gene-level information to detect and impute dropout events.
• Utilization of Unaffected Gene Expression: Borrows unaffected gene expression data from similar (analogous) cells to impute missing values while retaining inherent heterogeneity across cell types.
• Statistical Framework: Applies a statistical framework that automatically identifies dropout events by analyzing gene expression levels and their variations across similar cells and genes.
• Performance Evaluation: Demonstrated effectiveness on simulated and real datasets, improving the accuracy of downstream analyses compared with state-of-the-art imputation methods.

Scientific Applications:

• scRNA-seq Data Quality Improvement: Recovers missing expression values to enhance the quality and reliability of single-cell RNA-seq datasets.
• Preservation of Cellular Heterogeneity: Maintains heterogeneity of gene expression profiles across cells for accurate biological interpretation.
• Cell Diversity and Type Identification: Supports exploration of cell diversity and identification of distinct cell types.
• Downstream Analysis Enhancement: Improves accuracy of clustering, visualization, and differential expression analysis on imputed scRNA-seq data.

Methodology:

Identification of dropout events using statistical analysis of gene expression levels and their variations across similar cells and genes; block imputation by borrowing unaffected gene expression data from analogous cells to fill missing values.