Summary
A bioRxiv preprint reports that oversampling closely related prokaryotic strains makes genome collections less informative than their raw counts suggest. Applying the authors’ PhyloThin method to 402 species, the study estimates that the effective sample count is 41% below the actual count.
A bioRxiv preprint by Hannah Götsch and Franz Baumdicker reports that prokaryotic genome collections contain widespread and increasing sampling bias. Applying their method, PhyloThin, to 402 species in the NCBI genome database, the authors estimate that the effective number of samples is 41% lower than the actual number of genomes counted.
The difference matters because many genomes in a collection may come from closely related strains. A large raw count can therefore give a misleading impression of how broadly a species has been sampled. The authors say this uneven coverage affects estimates of gene frequencies and the amount of genomic diversity represented in databases.
How PhyloThin assesses sampling
Prokaryotes include bacteria and archaea. Their genomes are used to study microbial diversity and the distribution of traits, including antibiotic resistance. But public collections combine data generated by studies with different aims. The preprint says this can lead to particular groups of closely related strains being sampled repeatedly.
PhyloThin assesses sampling bias using phylogenetic relationships: information about how organisms are related through evolution. It compares the relationships among sampled genomes with what would be expected under random sampling. The authors say this approach accounts for sample size, unlike dereplication and other data-reduction methods, which are commonly used to reduce redundancy in genome collections.
The resulting estimate of an “effective” sample count is intended to describe how much useful sampling a collection provides, rather than simply how many genomes it contains. Across the 402 species analysed, the estimated effective count was 41% below the actual count. The authors also report substantial variation in oversampling between species, so the same raw genome count need not mean the same level of coverage for different organisms.
Implications for diversity and gene estimates
When some closely related strains are overrepresented, their genes can carry too much weight in estimates of how common a gene is. The authors report that accounting for oversampling reveals greater genomic diversity and more genes awaiting discovery than the existing data suggest. They say improved gene-frequency estimates could also matter for pathogen surveillance and searches for new genetic functions.
The findings make raw genome totals an incomplete measure of how well a species has been characterised. A collection with many genomes may still provide limited coverage of the diversity within a species if those genomes are concentrated among related strains. The authors describe sampling bias as widespread and increasing in recent years; the supplied preprint abstract does not provide the time period or species-level estimates behind that trend.
This work is a bioRxiv preprint, and its findings have not yet been peer reviewed. The authors identify the PhyloThin code repository on GitHub.