Shotgun metagenomic analysis of the human associated microbiome provides a rich set of microbial features for prediction and biomarker discovery in the context of human diseases and health conditions. However, the use of such high-resolution microbial features presents new challenges, and validated computational tools for learning tasks are lacking. Moreover, classification rules have scarcely been validated in independent studies, posing questions about the generality and generalization of disease-predictive models across cohorts. In this paper, we comprehensively assess approaches to metagenomics-based prediction tasks and for quantitative assessment of the strength of potential microbiome-phenotype associations. We develop a computational framework for prediction tasks using quantitative microbiome profiles, including species-level relative abundances and presence of strain-specific markers. A comprehensive meta-analysis, with particular emphasis on generalization across cohorts, was performed in a collection of 2424 publicly available metagenomic samples from eight large-scale studies. Cross-validation revealed good disease-prediction capabilities, which were in general improved by feature selection and use of strain-specific markers instead of species-level taxonomic abundance. In cross-study analysis, models transferred between studies were in some cases less accurate than models tested by within-study cross-validation. Interestingly, the addition of healthy (control) samples from other studies to training sets improved disease prediction capabilities. Some microbial species (most notably Streptococcus anginosus) seem to characterize general dysbiotic states of the microbiome rather than connections with a specific disease. Our results in modelling features of the “healthy” microbiome can be considered a first step toward defining general microbial dysbiosis. The software framework, microbiome profiles, and metadata for thousands of samples are publicly available at http://segatalab.cibio.unitn.it/tools/metaml.

No abstract

The human microbiome has emerged as a key aspect of human biology and has been implicated in many etiologies. Shotgun metagenomic sequencing is the most high-resolution approach available to study taxonomic composition and functional potential of the human microbiome, and an increasing amount of published data are available for re-use. These public data resources allow the possibility of rapid, inexpensive hypothesis testing for specific diseases and environmental niches, and meta-analysis across multiple related studies. However, several factors prevent the research community from taking full advantage of these public resources. Barriers include the substantial investments of time, computational resources, and specialized bioinformatic expertise required to convert them to analyzable form, and inconsistencies in annotation and formatting between individual studies.To overcome these challenges, we developed the curatedMetagenomicData data package (described at https://waldronlab.github.io/curatedMetagenomicData/) for distribution through the Bioconductor 1 ExperimentHub platform (see Supplementary Methods). curatedMetagenomicData provides highly curated and uniformly processed human microbiome data including bacterial, fungal, archaeal, and viral taxonomic abundances, in addition to quantitative metabolic functional profiles and standardized per-participant metadata. Data resources are accessible with a minimum of bioinformatic knowledge, while integration with the R/Bioconductor environment allows full flexibility for biologists, clinicians, epidemiologists, or statisticians to perform novel analyses and methodological development. We produced these resources by (i) downloading the raw sequencing data, (ii) processing it through the MetaPhlAn2 2 and HUMAnN2 3 pipelines, (iii) manually curating sample and study information, (iv) creating a pipeline to document and represent the above results as integrative Bioconductor objects, and (v) working with the Bioconductor core . CC-BY-NC 4.0 International license not peer-reviewed) is the author/funder. It is made available under a The copyright holder for this preprint (which was . http://dx

Background COVID‐19 pandemic is a global health crisis. Very few studies have reported association between obesity and severity of COVID‐19. In this meta‐analysis, we assessed the association of obesity and outcomes in COVID‐19 hospitalized patients. Methods Data from observational studies describing the obesity or body mass index (BMI) and outcomes of COVID‐19 hospitalized patients from December 1, 2019, to August 15, 2020, was extracted following PRISMA guidelines with a consensus of two independent reviewers. Adverse outcomes defined as intensive care units (ICU), oxygen saturation <90%, invasive mechanical ventilation (IMV), severe disease and in‐hospital mortality. The odds ratio (OR) and 95% confidence interval (95%CI) were obtained and forest plots were created using random‐effects models. Results A total of 10 studies with 10,233 confirmed COVID‐19 patients were included. The overall prevalence of obesity in our study was 33.9% (3473/10,233). In meta‐analysis, COVID‐19 patient with obesity had higher odds of poor outcomes compared to better outcomes with a pooled OR of 1.88 (95%CI:1.25–2.80; p=0.002), with 86% heterogeneity between studies (p<0.00001). Conclusion Our study suggests a significant association between obesity and COVID‐19 severity and poor outcomes. Our results findings may have important suggestions for the clinical management and future research of obesity and COVID‐19. This article is protected by copyright. All rights reserved.