Table 2 Good practices in microbiome studies

Design considerations

• The bacterial “universal” primers show bias in PCR amplification of certain taxa, and the use of different regions of the 16S rRNA gene in different studies (as seen for the lower airways in Table 1) affects inter-study comparisons (e.g. [21, 22])

Sample collection

• Make sure that collecting and storage vessels are not needlessly subjected to potential contamination either by body contact or air exposure.

DNA extraction

• Perform all activities (as long as it is practical to do so) under a hood with air filtering.

• Samples should be randomized before DNA extraction so that batch effects are minimized when groups of interest (case vs. control, age, sex, treatment, etc.) are compared. Varying contamination in kit lots and laboratory reagents can create artificial differences between groups during statistical analysis if the samples are not handled in randomized fashion (see [7] for an excellent and clear example).

• Be generous with controls. Every batch of samples being isolated at the same time should include one kit “blank” (control) to which no sample material is added but which undergoes the same process of DNA extraction and sequencing as the “real” samples. This serves the purpose of controlling for both contamination present in the kit and contamination introduced during the extraction process.

• Keep records of which kit lot was used for DNA extraction of which sample, and don’t mix reagents from different kit boxes, even if from the same lot.

• Use kits with bead-beating to increase the chances that taxa with thicker cell walls will be properly lysed and that taxonomic representation biases will be avoided as much as possible.

• Ensure that all samples from the same project are handled in the same way following a common protocol, each individual step preferably executed by the same person.

PCR

• When possible, work in a PCR clean room.

• Sequence a PCR master mix “blank” (control) for each different master mix aliquot. Do not add any template. Master mix controls serve the purpose of detecting potential contamination in PCR reagents (already present or accidentally introduced during preparation) every time a new master mix is prepared.

• Use PCR replicates to minimize PCR bias (uneven PCR amplification, lack of reaction effectiveness) and to detect the diversity present in samples as thoroughly as possible [113]. A minimum of two replicate reactions should be prepared per sample.

Sequencing

• If possible, sequence a mock community prepared from genomic DNA from known isolates. Since the sequence composition of this community is known, it can be used to identify contamination effects and sequencing errors in the target samples.