.abstract img { width:300px !important; height:auto; display:block; text-align:center; margin-top:10px } .abstract { overflow-x:scroll } .abstract table { width:100%; display:block; border:hidden; border-collapse: collapse; margin-top:10px } .abstract td, th { border-top: 1px solid #ddd; padding: 4px 8px; } .abstract tbody tr:nth-child(even) td { background-color: #efefef; } .abstract a { overflow-wrap: break-word; word-wrap: break-word; }
A4711 - Sequencing Vapor - Differentiating Something from Nothing When Sequencing Ultra Low-Biomass Samples
Author Block: J. Erb-Downward1, N. R. Falkowski1, R. A. McDonald1, J. L. Curtis2, B. Foxman3, J. C. D'Souza3, S. Adar3; 1Internal Internal Medicine-Pulmonary and Critical Care Medicine, Univ of Michigan Hlth System, Ann Arbor, MI, United States, 2Internal Medicine, Univ of Michigan Hlth System, Ann Arbor, MI, United States, 3Epidemiology, University of Michigan, Ann Arbor, MI, United States.
RATIONALE: The recent explosion of sequencing-based explorations has identified bacterial communities in samples once thought to be sterile. Although important findings have come from these studies, especially characterization of the human lung microbiome in health, there are obvious concerns when sequencing ultra-low biomass samples that contain bacterial signal at or less than the level of the negative controls. In 2014, Salter S. et.al. (BMC Biol. 2014;12:87. doi: 10.1186/s12915-014-0087-z) demonstrated that free bacterial DNA in the kits used for DNA isolation (frequently called the kitome or contaminome) could produce a reagent-specific signal that significantly altered the community profile of a sample. Here, we show that the kitome is only one component of systemic noise. Moreover, we demonstrate methods that address the question: in the ultra-low biomass setting, how does one know if they are sequencing nothing or something?
METHODS: We analyzed serial dilutions of thee defined sample types (P. aeruginosa, P. fluorescens, a manufactured mock community), and exhaled breath condensates from 8 healthy human volunteers), using MiSeq dual-barcode amplicon sequencing of the V4 region of the bacterial 16S rRNA gene. To assess detection of reagent blanks, we used multiple barcodes on the same sample; to identify kit contaminants, relevant negative controls from each of the matched DNA extraction kits were used. We validated results using qPCR. Sequences were processed in mothur and analyzed in R.
RESULTS: As concentrations of bacterial DNA decreased, an increasingly diverse "bacterial community" emerged in the first three bacterial sample types. A portion of this community, but not all, could be traced to reagent contamination. Replicate sequencing of samples using multiple barcodes per sample revealed that much of the non-kitome signal differed between replicates. Utilizing the geometric mean from multiple samples helped to identify both specific signal (from samples or reagents) and stochastic noise. We next used this methodology for ultra-low bacterial biomass samples to demonstrate that the observed bacterial community of exhaled breath condensate from healthy adults was not robust across repeated measures.
CONCLUSION: To ensure the veracity of bacterial community structure in ultra-low bacterial biomass samples (those at or below the level of controls), it is necessary to replicate sequences utilizing multiple barcodes to separate stochastic noise from real signal. Using this methodology and current sequencing technologies, we were unable to detect a bacterial community in ultra-low biomass exhaled breath condensate samples from 8 healthy subjects that was distinct from reagents or systemic noise.