coSMicQC

coSMicQC provides quality control for single-cell image-based profiling by identifying and filtering technical outliers to improve data quality and phenotype prediction.

Key Features:

• Systematic filtering: Supports systematic, human-in-the-loop filtering of low-quality single cells.
• Interactive threshold setting: Provides an interactive interface for setting thresholds and flagging technical outliers.
• Jupyter compatibility: Offers a Jupyter-compatible user interface for interactive analysis.
• Workflow integration: Integrates into standard image-based profiling protocols and existing workflows.
• Outlier detection: Detects technical outliers including mycoplasma contamination.
• Comparative performance: Outperforms PyOD at single-cell phenotype classification.

Scientific Applications:

• Assay optimization: Optimizes assay conditions for image-based profiling experiments.
• Contamination detection: Detects mycoplasma contamination in cell cultures.
• Drug screening rescue: Rescues lead compounds in large-scale drug screens that would otherwise be missed due to technical artifacts.
• Phenotype classification: Improves single-cell phenotype classification by removing low-quality cells.
• Heterogeneity analysis: Enables single-cell image-based profiling to reveal signals of phenotypic heterogeneity.

Methodology:

coSMicQC uses systematic, human-in-the-loop filtering with interactive threshold setting to flag technical outliers in single-cell image-based profiling data.