Abstract
High-throughput amplicon sequencing (HTAS) of conserved DNA regions is a powerful technique to characterize microbial communities. Recently, spike-in mock communities have been used to measure accuracy of sequencing platforms and data analysis pipelines. To assess the ability of sequencing platforms and data processing pipelines using fungal internal transcribed spacer (ITS) amplicons, we created two ITS spike-in control mock communities composed of cloned DNA in plasmids: a biological mock community, consisting of ITS sequences from fungal taxa, and a synthetic mock community (SynMock), consisting of non-biological ITS-like sequences. Using these spike-in controls we show that: (1) a non-biological synthetic control (e.g., SynMock) is the best solution for parameterizing bioinformatics pipelines, (2) pre-clustering steps for variable length amplicons are critically important, (3) a major source of bias is attributed to the initial polymerase chain reaction (PCR) and thus HTAS read abundances are typically not representative of starting values. We developed AMPtk, a versatile software solution equipped to deal with variable length amplicons and quality filter HTAS data based on spike-in controls. While we describe herein a non-biological SynMock community for ITS sequences, the concept and AMPtk software can be widely applied to any HTAS dataset to improve data quality.