Dynamics of gut bacteriophages in critically ill patients

Ho, Siu Fung ORCID: 0000-0001-9491-6498 (2024). Dynamics of gut bacteriophages in critically ill patients. University of Birmingham. Ph.D.

Preview

Ho2024PhD.pdf
Text - Accepted Version
Available under License All rights reserved.
Download (30MB) | Preview

Abstract

Phages are vital members of the gut microbiome and play a significant role in shaping their bacterial host community via predation, lysogeny, and transduction. While previous studies have identified that certain alterations in gut phage composition are associated with several diverse health conditions, the specific dynamics governing phage-host interactions remain largely unknown. Metagenomics and viromics have allowed for the direct identification and analysis of gut phages, regardless of their host’s cultivability. However, this process remains complex with significant challenges the scientific community has yet to definitively overcome.
In this study, we benchmarked ten commonly used metagenomic viral identification tools, and provided valuable metrics and insights for other researchers to use and build on. By using a mixture of artificial datasets and mock communities we were able to estimate the recall, precision, and biases of these tools. Overall, we concluded that Kraken2 should be considered when researchers are trying to identify previously characterised phages. When novel phage detection is required we suggest that Kraken2 is used in combination with machine learning based tools such as VirSorter2 and DeepVirFinder.
Subsequently, we identified viral operational taxonomic units (vOTUs) from a metagenomic sequencing set obtained by a previous study that investigated the gut microbiome of critically ill patients. We were able to define a straightforward yet robust criteria for microbial domination in the gut microbiota and use this to identify the dominant microbe in each patient, including one that had been discovered since the publication of the original study. The relative abundance of dominant microbes and vOTUs predicted to infect them had a near-perfect correlation. This correlation was not affected by the introduction of a time lag, leading us to hypothesise that the majority of our vOTUs originated from integrated prophages. We were also able to detect VFs and ARGs encoded by our vOTUs, although we were unable to determine whether these vOTUs were actively replicating or dormant.
We then built on this by experimentally removing bacterial hosts from stool samples collected from critically ill patients from Queen Elizabeth Hospital Birmingham. in order to long-read sequence the free viral fraction alone. In contrast with the metagenomic analyses, the dominating microbial taxa did not overlap with the most abundant hosts we predicted for our vOTUs, This observation lends support to the "piggyback-the-winner" model of gut phage dynamics. We identified two vOTUs that encoded \(cfxA\) beta-lactamases, including one that contained an ancestral form of the Tn\(4555\) transposon, while the second vOTU likely stemmed from a misassembly.
The insights gleaned from this work provide a framework for evaluating the performance of viral identification tools, as well as the advantages and limitations of utilising metagenomic versus viromic sequencing to explore the dynamics of the gut phageome.

Type of Work:

Thesis (Doctorates > Ph.D.)

Award Type:

Doctorates > Ph.D.

Supervisor(s):