Beyond Biodiversity: Incorporating Uncertainty Into Metabarcoding Data for Improved Inference of Ecological...

Area of Science:

• Ecology
• Genomics
• Bioinformatics

Background:

• Environmental DNA (eDNA) metabarcoding is a rapidly advancing technique for biodiversity assessment.
• eDNA data present unique challenges including complexity, sparsity, and compositional nature, stemming from sampling and sequencing uncertainties.
• Existing analytical methods struggle to fully account for these biases and the inherent uncertainty in eDNA marker gene data.

Purpose of the Study:

• To introduce MAMBO (Metabarcoding Analysis using Modeled Bayesian Occurrences), a novel computational tool.
• To provide a method for interpreting eDNA metabarcoding data that explicitly models uncertainty.
• To enable the linkage of different taxonomic groups through correlated marker gene analysis.

Main Methods:

• MAMBO simulates in silico replication to model uncertainty across the eDNA workflow.
• It employs Bayesian regression to correlate sequence count data from two different marker gene sets.
• The approach models uncertainty from patchy sampling, PCR amplification biases, and sequencing depth variations.

Main Results:

• MAMBO overcomes limitations of traditional correlational network analyses for eDNA data.
• The tool effectively models uncertainty inherent in metabarcoding sequence data.
• It facilitates the robust statistical analysis and interpretation of complex eDNA datasets.

Conclusions:

• MAMBO provides a robust framework for analyzing eDNA metabarcoding data, accounting for critical uncertainties.
• This approach enables more reliable correlation of different marker gene assays, linking diverse taxonomic groups.
• MAMBO offers new opportunities for gaining ecological insights into biodiversity patterns across space and time.