Towards standards for human fecal sample processing in metagenomic studies

Abstract

Technical variation in metagenomic analysis must be minimized to confidently assess the contributions of microbiota to human health. Here we tested 21 representative DNA extraction protocols on the same fecal samples and quantified differences in observed microbial community composition. We compared them with differences due to library preparation and sample storage, which we contrasted with observed biological variation within the same specimen or within an individual over time. We found that DNA extraction had the largest effect on the outcome of metagenomic analysis. To rank DNA extraction protocols, we considered resulting DNA quantity and quality, and we ascertained biases in estimates of community diversity and the ratio between Gram-positive and Gram-negative bacteria. We recommend a standardized DNA extraction method for human fecal samples, for which transferability across labs was established and which was further benchmarked using a mock community of known composition. Its adoption will improve comparability of human gut microbiome studies and facilitate meta-analyses.

Consistent and correctable bias in metagenomic sequencing experiments

“…Bias arises because each step in an experimental MGS workflow preferentially measures (i.e., preserves, extracts, amplifies, sequences, or bioinformatically identifies) some taxa over others (Brooks, 2016;Hugerth and Andersson, 2017;Pollock et al, 2018). For example, bacterial species differ in how easily they are lysed and therefore how much DNA they yield during DNA extraction (Morgan et al, 2010;Costea et al, 2017), and they differ in their number of 16S rRNA gene copies and thus how much PCR product we expect to obtain per cell (Kembel et al, 2012). Most sources of bias are protocol-dependent: Different PCR primers preferentially amplify different sets of taxa (Sipos et al, 2007), different extraction protocols can produce 10-fold or greater differences in the measured proportion of a taxon from the same sample (Costea et al, 2017), and almost every choice in an MGS experiment has been implicated as contributing to bias (Hugerth and Andersson, 2017;Sinha et al, 2017;Pollock et al, 2018).…”

“…For example, bacterial species differ in how easily they are lysed and therefore how much DNA they yield during DNA extraction (Morgan et al, 2010;Costea et al, 2017), and they differ in their number of 16S rRNA gene copies and thus how much PCR product we expect to obtain per cell (Kembel et al, 2012). Most sources of bias are protocol-dependent: Different PCR primers preferentially amplify different sets of taxa (Sipos et al, 2007), different extraction protocols can produce 10-fold or greater differences in the measured proportion of a taxon from the same sample (Costea et al, 2017), and almost every choice in an MGS experiment has been implicated as contributing to bias (Hugerth and Andersson, 2017;Sinha et al, 2017;Pollock et al, 2018). Every MGS experiment is biased to some degree, and measurements from different protocols are quantitatively incomparable (Nayfach and Pollard, 2016;Hiergeist et al, 2016;Mallick et al, 2017;Sinha et al, 2017;Gibbons et al, 2018).…”

Reproducibility: the search for microbiome standards

“…Variations resulting from different extraction methods were then compared, with differences attributed to library preparation and sample storage. The researchers demonstrated that DNA extraction had the largest effect on the outcome of metagenomic analysis [2]. …”

Clinical translation of microbe‐based therapies: Current clinical landscape and preclinical outlook

“…Therefore, it is difficult to determine the biological mechanism of action of FMTs from these clinical trials, which prevents an objective endpoint such as detectable presence of certain therapeutic strains. Large variations in the composition of the microbiota in both donors and patients and the lack of standard approaches for FMT sample processing, bacterial sequencing, and data analysis further complicate this challenge . As such, efforts to understand how factors such as phylum composition, data analysis, clinical endpoint time, antibiotic use, and microbe engraftment affect efficacy may provide mechanistic insight to better design and evaluate FMTs.…”

“…| 127 approaches for FMT sample processing, bacterial sequencing, and data analysis further complicate this challenge 38,58,59. As such, efforts to understand how factors such as phylum composition, data analysis, clinical endpoint time, antibiotic use, and microbe engraftment affect efficacy may provide mechanistic insight to better design and evaluate FMTs.…”