NEW YORK (GenomeWeb) – An international team of researchers has recommended a standardized DNA extraction method for use in studies of the human fecal microbiome.
Researchers rely on a variety of approaches to analyze the human fecal microbiome, each step of which may introduce variability, which has limited the interpretability of data generated through different protocols.
A team led by the European Molecular Biology Laboratory's Peer Bork focused on the DNA extraction step of fecal sample processing and tested more than 20 different protocols. As they reported today in Nature Biotechnology, they homed in on methods that have the highest extraction quality and that can be reproducibly used across labs. Based on their evaluations, they suggested that researchers in the field adopt the protocol they dubbed 'Q' to improve comparability.
"Taken together, our recommendations, if implemented across laboratories, may improve cross-study comparability and with this our ability to make meaningful inferences about the properties of the microbiome," the authors wrote in their paper.
Bork and his colleagues tested 21 DNA extraction methods coupled with shotgun metagenomic sequencing. To assess the methods, they generated multiple aliquots of two stool samples and sent them to 21 laboratories on three continents. The labs extracted DNA from the samples using the seven most common extraction kits as well as non-kit-based protocols. The labs then sent the DNA they extracted to Genoscope for centralized sequencing on the Illumina HiSeq 2000 platform.
The different extraction techniques lead to substantial technical variation, the researchers reported. The protocols generated differing amounts of DNA and differing levels of fragmented DNA. For instance, protocol 18 recovered 100 times more DNA than protocol 3 or protocol 12, while protocols 4, 10, and 12 generated highly fragmented DNA, but protocol 1 did not.
The researchers also examined whether the different approaches gave similar results in terms of the taxonomic and functional composition of the metagenomes. Most of the protocols provided similar species rankings, but some introduced batch effects at the species level.
Bork and his colleagues also found that some protocols yielded fewer numbers of Gram-positive bacteria — likely because they were unable to crack their cell walls to extract DNA. Because of this, they argued that diversity, including the presence of Gram-positive bacteria, could be used as an indicator of the performance of the DNA extraction step.
Bork and his colleagues selected the best-performing protocols for additional analysis. They sent the same stool samples to labs both familiar and unfamiliar with the chosen protocols to perform the extraction to test whether these protocols were reproducible. They also generated a mock bacterial community of 10 species not typically found in fecal samples and spiked those into eight additional stool samples for analysis.
All three methods tested were reproducible across locations, though protocol H had the lowest within-sample variation, the researchers said. However, protocol H also underestimated the amount of Gram-positive bacteria, as compared to the other two methods.
On the mock community samples, the three protocols performed similarly, the researchers reported, though protocol W nudged out the others.
However, the researchers noted that protocol W, owing to its reliance on phenol-chloroform, cannot be easily automated. Protocol Q, meanwhile, could recover a diverse microbial community and performed nearly as well as protocol W on an assay of accuracy, as it had a mean absolute quantification error of less than 0.5X.
Based on this, Bork and his colleagues concluded that protocol Q appears to the best overall approach and could serve as a benchmark against which new methods are compared. They further recommended that researchers use this protocol as a standardized extraction approach.
"Its adoption will improve comparability of human gut microbiome studies and facilitate meta-analyses," they wrote.