Run-to-Run Sequencing Variation Can Introduce Taxon-Specific Bias in the Evaluation of Fungal Microbiomes

The routine use of high-throughput sequencing to profile microbial communities necessitates improved protocols for detecting and adjusting for variation among sequencing runs for marker gene analysis. Although mock communities are widely used as a control among runs, the composition and diversity of mock communities, in most cases, are orders of magnitude lower than the actual samples. We demonstrated that replicated biological samples (“technical replicates”) are superior to a mock community in detecting variation and potential bias among sequencing runs. We present a case in which technical replicates of three soil samples were sequenced in three MiSeq runs containing samples from multiple experiments. The technical replicate samples revealed a potentially biased, outlier sequencing run, from which several Ascomycota taxa were substantially underestimated. Similar bias was seen in the other samples sequenced but was not detected using the mock community. Our study demonstrates that using technical replicates along with traditional mock communities provide additional quality control information and aid in detecting outlier sequencing runs.