Trait biases in microbial reference genomes.

Area of Science:

• Microbiology
• Genomics
• Bioinformatics

Background:

• Culturing techniques introduce biases, limiting our understanding of microbial diversity and traits.
• Existing genome databases heavily rely on cultured organisms, potentially skewing representation.

Purpose of the Study:

• To systematically examine biases in reference genome databases compared to natural microbial diversity.
• To quantify gene distribution differences between culture-independent metagenome-assembled genomes (MAGs) and culture-based RefSeq genomes.

Main Methods:

• Utilized 116,884 metagenome-assembled genomes (MAGs) from global surveys as a culture-independent dataset.
• Compared the prevalence of 12,454 KEGG orthologs (gene traits) in MAGs versus the RefSeq database.
• Employed statistical modeling to assess gene representation biases, controlling for environmental factors.

Main Results:

• The RefSeq database exhibits significant biases, under- or over-representing the majority of examined genes found in nature.
• Found disparities in gene distribution between MAGs and RefSeq, highlighting culture-driven biases.
• Identified conditional probabilities of species representation in RefSeq based on genetic repertoire.

Conclusions:

• Reference genome databases like RefSeq are not fully representative of natural microbial genetic diversity.
• The study provides a valuable resource for understanding and correcting gene-specific biases in microbial genomics.
• Highlights the need for culture-independent data to accurately characterize microbial traits.