The accuracy of absolute differential abundance analysis from relative count data

Abstract

Concerns have been raised about the use of relative abundance data derived from next generation sequencing as a proxy for absolute abundances. For example, in the differential abundance setting, compositional effects in relative abundance data may give rise to spurious differences (false positives) when considered from the absolute perspective. In practice however, relative abundances are often transformed by renormalization strategies intended to compensate for these effects and the scope of the practical problem remains unclear. We used simulated data to explore the consistency of differential abundance calling on renormalized relative abundances versus absolute abundances and find that, while overall consistency is high, with a median sensitivity (true positive rates) of 0.91 and specificity (1—false positive rates) of 0.89, consistency can be much lower where there is widespread change in the abundance of features across conditions. We confirm these findings on a large number of real data sets drawn from 16S metabarcoding, expression array, bulk RNA-seq, and single-cell RNA-seq experiments, where data sets with the greatest change between experimental conditions are also those with the highest false positive rates. Finally, we evaluate the predictive utility of summary features of relative abundance data themselves. Estimates of sparsity and the prevalence of feature-level change in relative abundance data give reasonable predictions of discrepancy in differential abundance calling in simulated data and can provide useful bounds for worst-case outcomes in real data.

Novel oil-associated bacteria in Arctic seawater exposed to different nutrient biostimulation regimes

“…Comparing cell density-corrected microbial community structure in oil-contaminated and non-contaminated microcosms using differential abundance analysis (48–50) identified a total of 41 microbial markers at the ASV level associated with crude oil amendment. In microcosms amended with crude oil in the presence of high nutrient concentrations, 22 microbial markers were identified, including bacterial taxa that have been previously described as hydrocarbonoclastic such as Oleispira , Algibacter , Shewanella , Neptunomonas , Halomonas and Pseudomonas (9, 10, 12, 14, 15, 51, 52) (Figure 3A).…”

“…Combining cell count and 16S rRNA gene diversity data for each microcosm enables total 16S rRNA gene abundances for different groups to be estimated. This eliminates the compositional effect of relative abundance data, which can otherwise increase false positive rates of microbial marker detection (48). In the case of hydrocarbonoclastic taxa, while many microbial lineages known to include hydrocarbon-degrading bacteria were enriched in the experiments performed here (Figure 2C; Figure 3) (9, 12, 51) , the microbiomeMarker pipeline also identified less well-understood genera Halarcobacter , Candidatus Pseudothioglobus and Lacinutrix as being among the most abundant groups in cold Northwest Passage seawater exposed to crude oil.…”

Airborne eDNA captures three decades of ecosystem biodiversity

“…Using sequencing data to estimate abundance is considered less promising because read counts provide catch-per-unit-effort (CPUE) data and are always affected by saturation ( 29 ). As with traditional CPUE surveys, reads can vary proportionally with abundance but how often this holds true for empirical datasets is uncertain ( 30 ). To test for proportionality, we searched for traditional inventories with sufficient spatial and temporal overlap with the eDNA time series.…”

Beyond Normalization: Incorporating Scale Uncertainty in Microbiome and Gene Expression Analysis

“…This phenomenon builds on the broader reproducibility issues in the biomedical sciences (Ioannidis, 2005). Non-biological differences in sequencing depth between samples can substantially contribute to the occurrence of false positives (Gloor et al, 2017;Vandeputte et al, 2017;McGovern et al, 2023;Nixon et al, 2023;Roche and Mukherjee, 2022;Props et al, 2017).…”

“…Sequence count data (e.g., 16S rRNA-seq or RNA-seq data) are ubiquitous in modern biological research. Statistical methods used to analyze these data often fail to control rates of false positives (Hawinkel et al, 2019; Nixon et al, 2023; Roche and Mukherjee, 2022). This phenomenon builds on the broader reproducibility issues in the biomedical sciences (Ioannidis, 2005).…”