The Three Peaks Challenge and developing extraction methods suitable for long-read, ultra-deep stool metagenomics on the PromethION

Josh Quick from the University of Birmingham began by talking about the motivation for using Nanopore sequencing for metagenomics. Their aim is to apply these metagenomics to a number of clinical projects at the University of Birmingham including one evaluating faecal microbiome transplants for ulcerative colitis. Using long nanopore reads to obtain single contig assemblies from complex metagenomic samples has a number of benefits: 1. It enables phylogenetics on whole genomes rather than individual reads. 2. It allows linking of genes to chromosomes. 3. It enables identification of strain level variation.

Josh went onto introduce the eponymous ‘Josh Quick’ method which combines Sambrook method for DNA isolation with the Nanopore rapid sequencing kit enabling the sequencing of reads in the 100’s of Kb range to > 1Mbase. As a proof of principle he was able to extract E.coli, sequence and assemble the genome using the 8 longest reads from the run. The assembly took just 1.5 sec on a single core using miniasm.

To look at sensitivity and the ability to assemble genomes from mixed metagenomic populations Josh sequenced the even and log-distributed Zymo Mock microbial community standards. Sequencing was run on the GridION and PromethION, generating over 150Gb on the PromethION runs.  The proportion of sequenced bases from each organism correlated well with expected proportions in this initial test where the DNA was extracted with a bead beating method. When looking at the data from the log mock community all 10 organisms were detected even S,aureus present at 0.000089% (~1pg DNA) and 6/10 had enough coverage to assemble genomes. Meaning assembly was possible over 4 logs difference in abundance, and detection over 6 logs.

Josh continued his talk discussing the problems that can be had with hard to lyse organisms: gram-positive bacteria have tough cell walls; yeasts have capsules or chitin layers and spores have strong coats. Failing to lyse these organisms will lead to failure to detect them via sequencing methods. Bead beating a sample will break almost all cells but it will also severely shear the DNA reducing read lengths. Josh has devised the ‘The Three Peaks method’ in an effort to increase read lengths but also ensure identification of all the species present in the sample. The three peaks method involves three sequential stages of lysis: chemical lysis, enzymatic lysis (metapolyzyme) and mechanical bead beading. Supernatants are taken after each stage of extraction. This extraction method enables extraction of DNA from even hard to extract organisms but also means that not all of the DNA is subjected to harsh bead beading which fragments the DNA significantly. Showing the results of clinical faecal samples Josh showed he we was able to improve the read lengths using the three peaks method generating N50s of up to 16Kb.

At the end of Josh’s talk he presented some pre-release R10 nanopore data from PCR’d and native DNA from the Zymo mock community. Six of the genomes in the mix had consensus accuracies >Q40 with the R10 data and the data is now available on github: https://github.com/LomanLab/mockcommunity