Identification and Removal of Potential Contaminants in 16S rRNA Gene Sequence Datasets from Low Microbial Biomass Samples: An Example from the Mosquito Gut

Abstract

Background: The bacterial gut microbiota of the female mosquito influences numerous physiological processes, including vector competence. As a low-microbial-biomass ecosystem, mosquito gut tissue is prone to contamination from the laboratory environment and from reagents commonly used to dissect and/or isolate DNA from gut tissue. In this report, we analyze five 16S rRNA datasets, including new data obtained by us, to gain insight into the impact of potential contaminating sequences on the composition, diversity, and structure of the mosquito gut microbial community. Results: We present a clustering-free approach that, based on the relative abundance of amplicon sequence variants (ASVs) in gut and negative control samples , allowed for the identification of candidate contaminating sequences. Some of these sequences belong to bacterial taxa previously identified as common contaminants in metagenomic studies; they have also been identified as part of the mosquito core gut microbiota, with putative physiological relevance for the host. By using different relative abundance cutoffs, we show that contaminating sequences have a significant impact on gut microbiota diversity and structure.Conclusions: The approach presented here allows the identification and removal of purported contaminating sequences in datasets obtained from low-microbial biomass samples. While it was exemplified with the analysis of gut microbiota from mosquitos, it can easily extend to other datasets dealing with similar technical artifacts.

Altered Nasal Microbiota Composition Associated with Development of Polyserositis by Mycoplasma hyorhinis

“…Denoising, trimming and quality-based filtering was performed with the q2-dada2 plugin [ 29 ], which also includes a chimera detection and consequent removal. Moreover, to improve the quality of our dataset, unassigned taxa was filtered out after aligning the ASVs against the Greengenes reference database (vs. 13.8) [ 30 ], clustered at 88% identity with 65% of identity and over the 50% of query [ 31 ].…”