Chord

Chord implements an AdBoost-based ensemble in R to integrate multiple doublet detection algorithms and improve identification of doublets in single-cell RNA sequencing (scRNA-seq) datasets for downstream analyses.

Key Features:

• Implementation (R): Implemented as an R package for scRNA-seq doublet detection.
• Integration of Multiple Methods: Uses an ensemble strategy that combines outputs from existing doublet detection algorithms via the AdBoost algorithm.
• High Accuracy and Stability: Demonstrated higher accuracy and stability in identifying doublets across both real and synthetic datasets.
• Modular Architecture: Modular design that facilitates incorporation of additional doublet detection tools.

Scientific Applications:

• Single-cell genomics quality control: Identification and removal of doublets from scRNA-seq datasets to improve data quality.
• Clustering: Improves clustering accuracy by removing artifactual doublet cells.
• Differential expression analysis: Reduces confounding effects of doublets in differential expression analyses.
• Trajectory inference: Enhances trajectory and lineage inference by eliminating doublet-induced artifacts.

Methodology:

Chord applies an AdBoost ensemble machine learning algorithm to integrate outputs from multiple doublet detection methods.