0 avis
CroCoDeEL: accurate detection of cross-sample contaminations in metagenomic data
Archive ouverte
Edité par CCSD -
International audience. Metagenomic sequencing provides deep insights into microbial communities but is subject to various experimental biases such as cross-sample contamination where microbial contents from simultaneously processed samples are accidentally mixed. Although a critical issue that can potentially lead to erroneous conclusions, such contamination remains understudied. A few methods have already been proposed to detect it, but with multiple limitations, notably the lack of sensitivity. Here, we introduce CroCoDeEL, a tool based on a supervised pre-trained model that identifies specific patterns in taxonomic profiles associated with cross-sample contamination. Benchmarks across three public cohorts comprehensively curated by the authors revealed that CroCoDeEL identifies with high accuracy not only the contaminated samples but also their respective contamination sources, even at low rates (< 0.1%) if the sequencing depth allows it. Our work underlines the urgency to acknowledge and systematically address this phenomenon to ensure the robustness of studies based on metagenomic data.
Consulter en ligne
Suggestions
Du même auteur
