Accurate Predictions of Microbial Diversity Using Poissonization
Author Information
Author(s): Lladser Manuel E., Gouet Raúl, Reeder Jens
Primary Institution: Department of Applied Mathematics, University of Colorado, Boulder, Colorado, United States of America
Hypothesis
Can we accurately predict the fraction of microbial species that remain unsampled in a given environment?
Conclusion
The study presents a new statistical method that provides accurate predictions for the fraction of unsampled microbial species in various environments.
Supporting Evidence
- The Embedding algorithm provides conditionally unbiased predictors for unsampled microbial species.
- Predictions were tested against simulated environments based on human-gut and -hand microbiota datasets.
- The method allows for accurate predictions even with small sample sizes.
Takeaway
This study helps scientists understand how many different types of tiny living things, like bacteria, are in a sample, even if they haven't found them all yet.
Methodology
The study uses a new statistical method called the Embedding algorithm, which models microbial samples as draws from an urn with an unknown composition.
Limitations
The method may yield inconclusive predictions when sample sizes are fixed and may not apply to all datasets.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website