Gut microbiome diversity detected by high-coverage 16S and shotgun sequencing of paired stool and colon sample

Abstract

The gut microbiome has a fundamental role in human health and disease. However, studying the complex structure and function of the gut microbiome using next generation sequencing is challenging and prone to reproducibility problems. Here, we obtained cross-sectional colon biopsies and faecal samples from nine participants in our COLSCREEN study and sequenced them in high coverage using Illumina pair-end shotgun (for faecal samples) and IonTorrent 16S (for paired feces and colon biopsies) technologies. The metagenomes consisted of between 47 and 92 million reads per sample and the targeted sequencing covered more than 300 k reads per sample across seven hypervariable regions of the 16S gene. Our data is freely available and coupled with code for the presented metagenomic analysis using up-to-date bioinformatics algorithms. These results will add up to the informed insights into designing comprehensive microbiome analysis and also provide data for further testing for unambiguous gut microbiome analysis.

Fig. 1

Taxonomic classification of samples at family level. (a) Classification of shotgun samples using three different classifiers. (b) Classification of 16S sequences, split by region and source material, using DADA2 and IdTaxa.

Fig. 2

Ordination. Principal components analysis of the datasets after central log ratio transformations of the family-level classifications. (a) 16S data, where each sample data was stratified by region and source material. (b) Shotgun data, classified using Kraken2, Kaiju and MetaPhlAn2. (c) 16S data from faeces (only V4 region) and shotgun data (classified using Kraken2).

Fig. 3

Alpha diversity. Shannon index was calculated at different taxonomic levels (species, genus, phylum, top row) as classified by Kraken2 and functional (gene families: UniRef90, functional groups: KEGG orthogroups and metabolic pathways: MetaCyc, bottom row) levels as classified by HUMAnN2 by number of read pairs. Five random samples were created at each level.