Complementing 16S rRNA gene amplicon sequencing with estimates of total bacterial load to infer absolute species concentrations in the vaginal microbiome

Abstract

Whereas 16S rRNA gene amplicon sequencing quantifies relative abundances of bacterial taxa, variation in total bacterial load between samples restricts its ability to reflect absolute concentration of individual species. Quantitative PCR (qPCR) can quantify individual species, but it is not practical to develop a suite of qPCR assays for every bacterium present in a diverse sample. We analyzed 1320 samples from 20 women with a history of frequent bacterial vaginosis, who self-collected vaginal swabs daily over 60 days. We inferred bacterial concentrations by taking the product of species relative abundance (assessed by 16S rRNA gene amplicon sequencing) and total bacterial load (measured by broad-range 16S rRNA gene qPCR). Log10-converted inferred concentrations correlated with targeted qPCR (r = 0. 935, p<2.2e-16) for seven key bacterial species. The mean inferred concentration error varied across bacteria, with rarer bacterial vaginosis-associated bacteria associated with larger errors. 92% of errors >0.5 log10 occurred when relative abundance was <10%. Many errors occurred during early bacterial expansion or late contraction. When relative abundance of a species is >10%, inferred concentrations are reliable proxies for targeted qPCR. However, targeted qPCR is required to capture bacteria at low relative abundance, particularly with BV-associated bacteria during the early onset of bacterial vaginosis.

A multi-view model for relative and absolute microbial abundances

“…This reinforces that even a model designed to account for the distribution of varying efficiencies cannot accurately predict the efficiency of an individual taxon when only relative abundance data are available. Note that these findings corroborate existing literature: Boshier et al (2019) found that BVAB2 spp. is a high-efficiency taxon, and McLaren et al (2019) found that G. vaginalis is a low-efficiency taxon.…”

“…While we did not find a reference to estimator (4) Boshier et al (2019) recently validated this proposal using taxon-specific qPCR primers and found it to be "predictive of absolute concentration with certain key exceptions," such as certain taxa and low biomass (low total bacterial concentration: q j=1 µ ij ) samples. Bonk et al (2018) give an excellent overview of sources of discrepancies between qPCR and 16S sequencing data.…”

“…Empirically, we found that modeling the efficiency of the different technologies is critical for obtaining accurate estimates of microbial abundance. Boshier et al (2019) found that a naïve approach consistently overestimates the concentration of certain taxa by an order of magnitude (e.g. BVAB2 ).…”

“…observations with 16S relative abundance data. These data are additionally described by Boshier et al (2019). We use these data for illustrative purposes only, and did not incorporate correlation between samples into our analysis.…”

Intermittent Lactobacilli-containing Vaginal Probiotic or Metronidazole Use to Prevent Bacterial Vaginosis Recurrence: Safety and Preliminary Efficacy by Microscopy and Sequencing

“…This yielded 401 unique ASVs in 629 samples, mapping to species (n=255; 63.6%), genus (n=116; 28.9%), or higher taxonomic level (n=30; 7.5%). Concentrations in cells/μl per ASV per sample were estimated by multiplying the ASV-specific copy-normalized relative abundance by the sample-specific 16S rRNA gene copies concentration 33,34 . Heatmaps of the 20 most common ASVs by sample are shown in Figure S1A for relative abundances and Figure S1B for concentrations.…”