Thread by @pathogenomenick

.@Scalene introducing the difficulties of long-read metagenomic sample preparation from complex samples: harder to lyse than bags of cytoplasm like E. coli or human cells. #nanoporeconf

We've mainly favoured ultra-long libraries with rapid kit, but recently @DrT1973 has developed a bead-free ligation method that can get 100kb+ reads. Started to move to ligation method routinely as good balance of length and yield.

We've been practicing on mock communities from @ZymoResearch as they act as very useful positive controls for lab and bioinformatics. Zymo community consists of 3 Gram-, 3 Gram+ and 2 Fungi. #nanoporeconf

Important to preserve supernatant because some amount of lysis can occur with preservatives like DNA/RNA shield.

Presenting the recent paper from the group demonstrating sequencing mock community on GridION and PromethION: academic.oup.com/gigascience/ar…

PromethION sensitive down to a few hundred cells in a mixture up to 10^8 cells. Abundances as expected, but here we used bead beating. #nanoporeconf

But we need longer reads: aim to pick apart the extraction protocol.

First observation: spin columns significantly inferior to magnetic clean-ups (even with bead beating). #nanoporeconf

Next problem: lysis. Gram positive bacteria and fungi are particularly hard to lyse compared with many Gram negs. #nanoporeconf

One solution is metapolyzyme developed by Scott Tighe containing a six cocktail enzyme, aim to generate spheroplasts/protoplasts. #nanoporeconf

We've developed a workflow called the '3 peaks challenge': 1) chemical lysis, 2) enzymatic lysis and further 3) physical lysis with bead beating to aim to preserve long fragments and generate good representation. #nanoporeconf

It's not alright to ignore recalcitrant organisms if you want to properly study microbiomes. This protocol strikes a good balance.

Three peaks challenge named after Scarfield, Snowden and Ben Nevis. The challenge is to climb all those peaks in 24 hours!

3 Peaks method increases N50 on mock community from around 6kb to 25kb.

You can't ignore bead beating because metapolyzyme for example won't digest Cryptococcus neoformans.

We've started testing the 3 Peaks method on real clinical samples: ~26Gb on MinION, ~100Gb on PromethION with best N50 of 16.5kb.

3 Peaks refers to the three modes you should see on length distribution from the three separate lysis steps.

One last thing .... we've produced two R10 datasets (3 Peaks and native) that we are able to publish today.

Able to generate nanopore-only assemblies with >Q40 for 6 of the bacteria in the mixture using R10 data!

The R10 data and signal is available to download now on our mock community page!
lomanlab.github.io/mockcommunity/…

Storming talk from @Scalene, long read metagenomics is very exciting!