Benchmarking Metagenomic Classifiers on Simulated Ancient and Modern Metagenomic Data

Area of Science:

• Ancient DNA research
• Microbiome analysis
• Bioinformatics

Background:

• Ancient metagenomic sample analysis is hindered by DNA degradation.
• Existing metagenome profilers are primarily tested on modern or simulated modern samples.
• A comprehensive comparison of profiler performance across a degradation spectrum is lacking.

Purpose of the Study:

• To evaluate the performance of various metagenome profilers on simulated ancient DNA.
• To assess the impact of different DNA damage patterns (deamination, fragmentation, contamination) on profiler accuracy.
• To identify the strengths and weaknesses of different profiling approaches for ancient metagenomes.

Main Methods:

• Simulated metagenomes of human dental calculus microbiome were created with increasing DNA damage levels.
• Three classes of profilers (DNA-to-DNA, DNA-to-protein, DNA-to-marker) were assessed.
• Simulated deamination, fragmentation, and contamination were applied to mimic ancient DNA characteristics.

Main Results:

• Human and environmental contamination had the most significant negative impact on profiler performance compared to deamination and fragmentation.
• DNA-to-DNA profilers (e.g., Kraken2, Bracken) and DNA-to-marker profilers (e.g., MetaPhlAn4) demonstrated distinct performance characteristics.
• The effectiveness of profilers varied depending on the type and extent of DNA damage simulated.

Conclusions:

• Ancient DNA contamination is a critical factor affecting metagenome profiling accuracy.
• Combining DNA-to-DNA and DNA-to-marker approaches can enhance the reliability of ancient microbiome profiling.
• Further development and validation of profilers are needed for robust ancient metagenome analysis.