Assessing the comparability of different DNA extraction and amplification methods in gut microbial community profiling

Abstract

Automated, high-throughput technologies are becoming increasingly common in microbiome studies to decrease costs and increase efficiency. However, in microbiome studies, small differences in methodology – including storage conditions, wet lab methods, sequencing platforms and data analysis – can influence the reproducibility and comparability of data across studies. There has been limited testing of the effects of high-throughput methods, including microfluidic PCR technologies. In this paper, we compare two extraction methods (the QIAamp DNA Stool Mini Kit and the MoBio PowerSoil DNA Isolation kit), two taq polymerase enzymes (MyTaq HS Red Mix and Accustart II PCR ToughMix), two primer sets (V3–V4 and V4–V5) and two amplification methods (a common two-step PCR protocol and amplicon library preparation on the Fluidigm Access Array system that allows automated multiplexing of primers). Gut microbial community profiles were significantly affected by all variables. While there were no significant differences in alpha diversity measured between the two extraction methods, there was an effect of extraction method on community composition measured by unweighted UniFrac distances. Both amplification method and primers had a significant effect on both alpha diversity and community composition. The relative abundance of Actinobacteria was significantly lower when using the MoBio kit or Fluidigm amplification method, and the relative abundance of Firmicutes was lower when using the Qiagen kit. Microbial community profiles based on Fluidigm-generated amplicon libraries were not comparable to those generated with more commonly used methods. Researchers should carefully consider the limitations and biases that different extraction and amplification methods can introduce into their results. Additionally, more thorough benchmarking of automated and multiplexing methods is necessary to determine the magnitude of the potential trade-off between the quality and the quantity of data.