The use of taxon-specific reference databases compromises metagenomic classification.

Area of Science:

• Microbiology
• Bioinformatics
• Genomics

Background:

• Metagenomic studies often rely on bioinformatics tools to classify microbial communities.
• The HumanMycobiomeScan tool was developed to identify fungal taxa in metagenomic samples.
• Previous analyses using this tool characterized the human gut mycobiome of different populations.

Purpose of the Study:

• To evaluate the accuracy of taxon-specific reference databases in metagenomic classification.
• To demonstrate the potential for false-positive identifications using such databases.
• To propose a more robust approach for analyzing microbial eukaryotes in metagenomic data.

Main Methods:

• Re-analysis of a case study using a reference database containing amphibian and reptile genomes.
• Comparison of results from a fungal-only database versus a kingdom-agnostic database.
• Application of sequence read mapping against curated reference genomes.

Main Results:

• A fungal-specific database incorrectly identified herpetological taxa (turtles, frogs, snakes) as abundant in the human gut.
• Re-analysis with a kingdom-agnostic pipeline revealed a broader range of microbial eukaryotes but fewer specific fungal families.
• The study identified significant limitations in using taxon-specific databases for metagenomic classification.

Conclusions:

• Taxon-specific reference databases can yield high false-positive rates in metagenomic analysis.
• The use of curated, comprehensive databases encompassing all domains of life is crucial for accurate species profiling.
• This highlights the importance of selecting appropriate bioinformatics pipelines for microbiome research, especially for microbial eukaryotes.