Bayesian compositional regression with microbiome features via variational inference.

The microbiome plays a key role in the health of the human body. Interest often lies in finding features of the microbiome, alongside other covariates, which are associated with a phenotype of interest. One important property of microbiome data, which is often overlooked, is its compositionality as it can only provide information about the relative abundance of its constituting components. Typically, these proportions vary by several orders of magnitude in datasets of high dimensions. To address these challenges we develop a Bayesian hierarchical linear log-contrast model which is estimated by mean field Monte-Carlo co-ordinate ascent variational inference (CAVI-MC) and easily scales to high dimensional data. We use novel priors which account for the large differences in scale and constrained parameter space associated with the compositional covariates. A reversible jump Monte Carlo Markov chain guided by the data through univariate approximations of the variational posterior probability of inclusion, with proposal parameters informed by approximating variational densities via auxiliary parameters, is used to estimate intractable marginal expectations. We demonstrate that our proposed Bayesian method performs favourably against existing frequentist state of the art compositional data analysis methods. We then apply the CAVI-MC to the analysis of real data exploring the relationship of the gut microbiome to body mass index.

Microbial Signatures in COVID-19: Distinguishing Mild and Severe Disease via Gut Microbiota.

The COVID-19 pandemic, caused by the SARS-CoV-2 virus, has significantly impacted global healthcare, underscoring the importance of exploring the virus's effects on infected individuals beyond treatments and vaccines. Notably, recent findings suggest that SARS-CoV-2 can infect the gut, thereby altering the gut microbiota. This study aimed to analyze the gut microbiota composition differences between COVID-19 patients experiencing mild and severe symptoms. We conducted 16S rRNA metagenomic sequencing on fecal samples from 49 mild and 43 severe COVID-19 cases upon hospital admission. Our analysis identified a differential abundance of specific bacterial species associated with the severity of the disease. Severely affected patients showed an association with Enterococcus faecium, Akkermansia muciniphila, and others, while milder cases were linked to Faecalibacterium prausnitzii, Alistipes putredinis, Blautia faecis, and additional species. Furthermore, a network analysis using SPIEC-EASI indicated keystone taxa and highlighted structural differences in bacterial connectivity, with a notable disruption in the severe group. Our study highlights the diverse impacts of SARS-CoV-2 on the gut microbiome among both mild and severe COVID-19 patients, showcasing a spectrum of microbial responses to the virus. Importantly, these findings align, to some extent, with observations from other studies on COVID-19 gut microbiomes, despite variations in methodologies. The findings from this study, based on retrospective data, establish a foundation for future prospective research to confirm the role of the gut microbiome as a predictive biomarker for the severity of COVID-19.

16S rRNA gene sequencing data of the upper respiratory tract microbiome in the SARS-CoV-2 infected patients.

The SARS-CoV-2 pandemic is a big challenge for humanity. The COVID-19 severity differs significantly from patient to patient, and it is important to study the factors protecting from severe forms of the disease. Respiratory microbiota may influence the patient's susceptibility to infection and disease severity due to its ability to modulate the immune system response of the host organism. This data article describes the microbiome dataset from the upper respiratory tract of SARS-CoV-2 positive patients from Russia. This dataset reports the microbial community profile of 335 human nasopharyngeal swabs collected between 2020-05 and 2021-03 during the first and the second epidemic waves. Samples were collected from both inpatients and outpatients in 4 cities of the Russian Federation (Moscow, Kazan, Irkutsk, Nizhny Novgorod) and sequenced using the 16S rRNA gene amplicon sequencing of V3-V4 region. Data contains information about the patient such as age, sex, hospitalization status, percent of damaged lung tissue, oxygen saturation (SpO2), respiratory rate, need for supplemental oxygen, chest computer tomography severity score, SARS-CoV-2 lineage, and also information about smoking and comorbidities. The amplicon sequencing data were deposited at NCBI SRA as BioProject PRJNA751478.