Consistent and correctable bias in metagenomic sequencing experiments

Abstract

Marker-gene and metagenomic sequencing have profoundly expanded our ability to measure biological communities. But the measurements they provide differ from the truth, often dramatically, because these experiments are biased toward detecting some taxa over others. This experimental bias makes the taxon or gene abundances measured by different protocols quantitatively incomparable and can lead to spurious biological conclusions. We propose a mathematical model for how bias distorts community measurements based on the properties of real experiments. We validate this model with 16S rRNA gene and shotgun metagenomics data from defined bacterial communities. Our model better fits the experimental data despite being simpler than previous models. We illustrate how our model can be used to evaluate protocols, to understand the effect of bias on downstream statistical analyses, and to measure and correct bias given suitable calibration controls. These results illuminate new avenues toward truly quantitative and reproducible metagenomics measurements.

A multi-view model for relative and absolute microbial abundances

“…Given that the collection of qPCR data involves calibration (via a "standard curve") and 16S relative abundance data does not usually involve any calibration, we modeled the efficiency of the 16S data compared to the qPCR data, rather than the efficiency of the qPCR data compared to the 16S data. This is consistent with recent literature highlighting the lack of calibration and resulting bias of 16S data for estimating relative abundance (McLaren et al 2019). Our method can also be used with other technologies for obtaining absolute and relative abundance data.…”

“…, e q ), and e j is the efficiency of taxon j for being observed by the relative abundance technology compared to the absolute abundance technology. Our efficiency vector e plays the role of the "total protocol bias" parameter of McLaren et al (2019). We now discuss estimation of the parameters of this model, including the identifiability of the efficiencies e.…”

Current Trends in Diagnostics of Viral Infections of Unknown Etiology

“…Nucleic acids from a sample undergo virtually unbiased sequencing, i.e., with minimum prejudice towards specific organisms [47,48]; in theory, the method can be used to analyze a potentially unlimited range of targets [49,50]. Nevertheless, there is evidence that metagenomic sequencing is restricted by multiple pitfalls and biases, based on the pathogen's structure, extraction method, GC content, and other factors [51][52][53]. Therefore, there are five major limitations: (1) sequences of interest should share at least low identity with the analogous sequences in a reference genome, ensuring correct mapping; (2) the analytical complexity of obtained data often requires employment of a qualified bioinformatician and the use of a specialized computational infrastructure (including both hardware and software; (3) hidden experimental and methodological prejudices towards certain taxa; (4) defining the method's sensitivity and properly measuring it against a relevant reference; (5) dealing with the abundance of host-cell, bacterial and fungal nucleic acids.…”

Host genotype and colonist arrival order jointly govern plant microbiome composition and function

“…Each species was present at each concentration in 3 mock communities and all 9 mock communities had 3 species at each concentration. The resulting Sequencing read counts and the known proportional abundances of each species in each mock community were then used to estimate species-specific biases following the method of (McLaren et al, 2019). We found significant bias among the members of our mock communities (Figs.…”

“…A tractable approach to overcoming these limitations is to assemble synthetic microbial communities, where immigration history can be precisely controlled (Vorholt et al, 2017;Carlström et al, 2019). In addition, by using a simplified microbial community, with known membership, it is also possible to quantify and correct sequencing biases to more accurately assess community assembly outcomes (McLaren et al, 2019).…”