vaeda

vaeda identifies and scores doublets—technical artifacts that occur when two or more cells are tagged with the same barcode—in single-cell RNA sequencing (scRNA-seq) data to enable accurate downstream analyses.

Key Features:

• Variational Auto-Encoder (VAE) integration: Uses a variational auto-encoder framework to model complex scRNA-seq data distributions for distinguishing single cells from doublets.
• Positive-Unlabeled learning: Applies Positive-Unlabeled learning to refine quantitative doublet scores and produce binary doublet calls.
• Quantitative and binary outputs: Produces both continuous doublet scores and qualitative binary doublet annotations.
• Benchmarking performance: Evaluated on 16 benchmark datasets against seven existing methods, demonstrating competitive performance and outperforming other Python-based tools in the reported comparisons.
• Robustness across datasets: Demonstrated robustness in handling diverse scRNA-seq datasets.

Scientific Applications:

• Transcriptional heterogeneity studies: Ensures integrity of single-cell expression profiles for accurate analysis of cellular diversity.
• Developmental biology: Provides precise doublet annotation to support single-cell studies of developmental processes.
• Cancer genomics: Helps distinguish true tumor subpopulations from technical doublet artifacts in tumor heterogeneity research.

Methodology:

vaeda combines a variational auto-encoder framework with Positive-Unlabeled learning to model scRNA-seq data distributions and generate quantitative doublet scores along with binary calls.