Squeegee

Squeegee detects contaminant taxonomic signals in metagenomic sequencing datasets to identify laboratory- or kit-derived contamination, with emphasis on low-biomass samples.

Key Features:

• De novo contamination detection: Identifies contaminant sequences without requiring external reference negative controls by operating de novo on dataset-internal signals.
• Taxonomic "bread crumbs" hypothesis: Leverages the hypothesis that contaminants leave shared taxonomic signatures across different sample types to detect contamination.
• Pattern and similarity analysis: Analyzes patterns and similarities of taxonomic signals across metagenomic datasets to distinguish contaminants from genuine community members.
• Low-biomass focus: Specifically applied to low-biomass metagenomic samples where contamination can disproportionately affect results.
• Validation on simulated and real data: Demonstrated precision by recovering known contaminants in simulated datasets and by comparison to experimental negative-control data in real datasets.
• Large-scale retrospective analysis: Applied to 749 Human Microbiome Project metagenomic datasets to reveal previously unreported kit contamination.

Scientific Applications:

• Host-associated microbiome studies: Detection and filtering of contaminants to improve accuracy of host-associated metagenomic analyses.
• Low-biomass sample analysis: Identifying contamination sources in environments where true microbial signal is weak relative to contaminant signal.
• Retrospective screening of public datasets: Discovery of kit-derived and laboratory contaminants in large-scale public metagenomic collections such as the Human Microbiome Project.

Methodology:

Performs de novo identification of contaminant taxa by analyzing taxonomic signal patterns and similarities across metagenomic datasets, leveraging shared taxonomic "bread crumbs" to detect contamination without negative controls.