Long-Read Metagenomics to Retrieve High-Quality Metagenome-Assembled Genomes from Canine Feces

Anna Cuscó (Vetgenomics, Spain) began her talk by discussing the approaches available to investigate microbiomes: amplicon-based or metagenomics-based. Amplicon-based strategies are generally employed when low amounts of sample are available; however, this method is limited to taxonomic information, and microbes other than bacteria are missed. When higher biomass samples are available, metagenomics is possible, enabling the generation of metagenome-assembled genomes (MAGs), antimicrobial resistance (AMR) and virulence profiling, assessment of the functional potential of the community, and detection of all microbes present.

Anna and her team used nanopore metagenomic sequencing to characterise the microbiome of dog faecal samples. DNA was extracted using two methods: the Zymobiomics DNA Miniprep kit, a bead beating-based method, or via Quick-DNA HMW MagBead, without bead-beating, to obtain high molecular weight DNA. Libraries were prepared using the Ligation Sequencing Kit and sequenced on the MinION device. Raw read correction was performed using the tool Canu, then assembly using Flye and polishing via Medaka. Medium- and high-quality MAGs were then extracted from this dataset, further corrected using DIAMOND and MEGAN-LR (Arumugam et al.) to improve completeness, annotated with PROKKA, and their taxonomy/novelty assessed with GTDB-tk.

Displaying bandage plots for high-quality MAGs assembled from the two libraries, Anna demonstrated that those assembled from the HMW DNA library produced larger contigs than the DNA extracted via bead-beating, with some MAGs representing closed, circular genomes. The HMW data also produced more high-quality MAGs. Next, they merged the two datasets to maximise consensus accuracy, experimenting with different proportions of data from each run to produce eight high-quality MAGs. Two of these, Enterococcus hirae and Blautia argii, displayed >99.6% identity with reference genomes, the latter of which had previously been identified in canine faeces, validating their approach. Four MAGs matched existing reference genomes for bacteria present in the human gut microbiome; however, where the reference genomes were comprised of 87-212 contigs, each of the MAGs produced by Anna and her team were complete, improving on the quality of the references. Their MAGs also contained all the rRNAs and a higher number of tRNAs. Anna highlighted how these MAGs meet the high-quality criteria required for MIMAG.

The two remaining MAGs represented potential novel species: Succinivibrio sp. and Sutterella sp. Analysis of the 16S rRNA of the former revealed identity with the 16S sequence of one uncultured clone from a wolf gastrointestinal microbiome. The latter showed >99% identity with the 16S sequence of Sutterella stecoricanis, which has been identified in canine faeces, but represented the first genome assembly for this species. Functional analysis of Succinivibrio sp. revealed abundant ‘mobilome’ functions compared with representatives from the human gastrointestinal microbiome. Anna and her team suspect that, rather than due to a biological reason, this abundance is a result of the use of long reads, which are able to capture the genomic context of mobile genetic elements; these cannot be spanned by the short reads used to assemble the other genomes in the comparison.