PeacoQC

PeacoQC performs automated quality control of cytometry data measuring numerous markers across millions of cells by identifying and removing low-quality events caused by anomalies such as clogs, speed changes, and slow sample uptake to preserve data integrity for downstream analyses.

Key Features:

• Density peak detection: Identifies density peaks within each channel of a cytometry dataset.
• Isolation tree assessment: Assesses events based on their position within an isolation tree structure.
• Mean absolute deviation scoring: Calculates mean absolute deviation distances from identified peaks to score events.
• Event filtering: Filters out low-quality events that could skew downstream analyses.
• Supported modalities: Validated on flow cytometry, mass cytometry, and spectral cytometry datasets.
• Benchmark performance: Achieved the highest median balanced accuracy compared to flowAI, flowClean, and flowCut while maintaining comparable running time.
• Scalability: Demonstrated improved scalability for handling large datasets.
• Parameter robustness: Tested on 16 public datasets with only one sample requiring further parameter optimization.
• Implementation: Implemented in R.

Scientific Applications:

• Cytometry data quality control: Automated cleaning of flow, mass, and spectral cytometry datasets to remove measurement artifacts.
• Preprocessing for downstream analyses: Reduces false discoveries by removing low-quality events prior to downstream analyses.
• Benchmarking of QC algorithms: Enables comparative evaluation of quality-control methods such as flowAI, flowClean, and flowCut.
• Parameter validation: Assesses robustness of QC parameters across public cytometry datasets.

Methodology:

PeacoQC detects density peaks per channel, assesses events by their position within an isolation tree structure, computes mean absolute deviation distances from those peaks, and filters events based on these scores.