From: Identifying biases and their potential solutions in human microbiome studies
Processing step | Recommended best practices |
---|---|
Sample collection | - Collect biological and technical replicates when possible. - Use the same collection device and manufacturer. - When possible, use the same collection personnel. - Use aseptic techniques during sample collection. - Make note of any variations during sample collection and include them in downstream analysis. |
Sample storage | - When possible freeze samples at −80°C immediately upon collection. - Exposure of samples to room temperature conditions should be minimized. - Preservatives should be used only when freezing of samples is infeasible (e.g., self-collected human samples). - If used, all samples should be stored in the same preservative. - Length of sample storage should be noted and included in downstream analysis. |
DNA extraction | - Extractions should be done using validated extraction kits or validated protocols such as those presented by the Earth Microbiome project or International Human Microbiome Standards. - All samples must be extracted with the same protocol. - Extraction batches should be noted and used as covariates in downstream analysis. - Extraction should include a mechanical lysis step (e.g., bead-beating). - Extraction should be done using aseptic techniques and a biological safety cabinet to reduce the amount of possible contamination. - A small pool of samples should be extracted during each extraction batch and sequenced to determine technical variation. - Blanks should be carried through extraction to sequencing (critical for low-biomass samples). |
PCR amplification | - Use high-fidelity polymerases. - Minimize the number of required PCR cycles (preferably max. 20–25). - Obtain as uniform amplification as possible for all samples. - Primers should be chosen based on the microbes of interest and whether the work is to be compared against previous literature. |
Metagenomic library construction | - Use equal amounts of template for library construction. - Avoid usage of discontinued Illumina Nextera XT kit. - Note mechanical sonication can cause minor biases toward high-GC content sequences. |
Marker gene bioinformatics | - Use denoising algorithms such as Deblur, DADA2, or UNOISE. - Use validated taxonomic classification methods such as QIIME2 feature classifiers or Kraken2. - Taxonomic classifiers should be consistent between comparison studies. - Use well-curated, up-to-date, comprehensive taxonomic databases such as SILVA, RDP, or NCBI. |
Metagenomic shotgun bioinformatics | - Referenced-based analysis:  ◦ Use DNA-based KMER taxonomic assignment such as Kraken2 or taxMaps.  ◦ Removal/filtering of low abundance taxa is recommended.  ◦ Employ phylogenetic marker-gene-based strategies when examining low abundance taxa. - Metagenomic assembly:  ◦ Use well-validated assembly methods such as MEGAHIT or metaSPAdes.  ◦ Use validated binning methods such as DASTool, MetaWrap, or MetaBat2. - For all methods, reference databases should be well-curated and up-to-date such as:  ◦ NCBI RefSeq  ◦ Genome Taxonomy Database |