SCRIBE

SCRIBE imputes dropout events and corrects batch effects in single-cell RNA sequencing (scRNA-seq) data to recover accurate cell-level gene expression while separating technical from biological variation.

Key Features:

• Simultaneous imputation and batch correction: Imputes dropout-induced zeros and corrects batch effects jointly in scRNA-seq datasets.
• Statistical modeling of technical versus biological variation: Employs a comprehensive statistical model that distinguishes technical noise from true biological signal.
• Modeling of dropout zero-inflation: Accounts for dropout events that produce inflated zero counts across single-cell measurements.
• Preservation of biological variation: Recovers cell-specific expression patterns while maintaining true biological differences across cells.
• Benchmark performance: Demonstrated superior imputation and batch-effect removal on real datasets compared with existing methods.
• R implementation and file-based inputs: Provided as an R script that accepts an expression file, a batch index file, a biological group index file, and an output file.

Scientific Applications:

• Expression recovery: Reconstruction of missing gene expression values in scRNA-seq data for accurate cell-level profiles.
• Batch-integrated analysis: Removal of batch effects to enable combined analysis across experimental batches.
• Cell-type clustering: Improved clustering of cells based on denoised expression profiles.
• Differential expression analysis: More reliable identification of differentially expressed genes after imputation and batch correction.
• Trajectory inference: Enhanced reconstruction of developmental or differentiation trajectories from single-cell data.

Methodology:

Applies a comprehensive statistical model that separates technical and biological variation and performs simultaneous dropout imputation and batch-effect correction in scRNA-seq data.