A couple of weeks ago a colleague asked me if we could do something on Nanopore sequencing for BBC Radio 4’s Today programme as it was due to broadcast from Aberystwyth. Thanks to prior experience with sequencing for the media, I said yes on the basis I could pick what we should sequence. Here I’ll summarize something of the process, outcome and implication.
Our brief was to extract DNA, prepare libraries, sequence and analyse soil metagenomes from two soil samples. This aligned well with the programme’s themes: nestling among serious and timely issues of student mental health, University funding and Brexit lay discussion ofthe implications of Brexit on food security. Soil health is an important aspect of that discussion. My intention was to shine a light on the typically hidden microbial diversity of soil as a serious-yet-not-so-serious way of highlighting the potential of portable metagenomics. We were to compare garden soil samples provided by the host, Justin Webb and from my Vice Chancellor.
What did we do?
In about ~15 minutes of sequencing per sample, we generated 70-74,000 raw reads which were basecalled and profiled using kaiju. Kaiju is an on-line and standalone taxonomic classifier that uses protein level matches between all potential reading frames in your DNA read and the reference database. We’ve found it to be fairly accurate when used with Nanopore data, and both quick and reliable in returning results. Important considerations for taking DNA sequencing to the flagship programme of a national radio station. In all, from sample to insight took ca. 2h35 from the programme’s start including DNA extraction, library prep, sequencing and analysis.
You can hear me talk about this at 1h44 and 2h57 here.
What did we learn?
In each sample, we could assign the reads to 6,369 taxa (Aberystwyth) and 6,461 (Justin Webb) from ca. 74,000 and 70,000 raw reads respectively. These range between species level classifications (as defined by Kaiju) to reads left assigned at phylum. This means Justin’s compost has marginally more detectable biodiversity than soil scant metres from a UNESCO Man and Biosphere Reserve. To my eye, this raises an important point about microbial biodiversity, its biogeography and its conservation. What we think we know from plants and animals doesn’t always fare well in translation.
Thanks to a 3 AM soil pH checkup last night I knew we had soil samples with pH 4.62 (Aberystwyth) and pH 5.57 (Justin Webb). There’s a litany of papers describing soil pH as a potential key driver of soil bacterial diversity, through a range of direct and indirect mechanisms. Here’s a fantastic example from Rob Griffiths at the Centre for Ecology and Hydrology. A map of the UK like no other. It was no surprise the Aberystwyth soil was about twice as rich in assignments to the Acidobacteria phylum as Justin’s compost. Notably, there were 60 discrete taxa within just one of the classes of Acidobacteria. There’s still a lot we don’t know about Acidobacteria. I think I would be oversimplifying if I just said they are good at tolerating acidic conditions, perhaps it is more to do with dealing well with other things that come in the wake of low pH.
On air I described Justin’s soil as being home to many “common or garden bacteria”. By this I mean there were many members of Alphaproteobacteria, specifically Bradyrhizobiales. In that regard his soil was home to some of the most prevalent bacteria in soils. Often the bane of low-biomass studies, here we were dealing with > 1.5 micrograms of soil DNA so it seems legit.
Justin seemed strangely preoccupied with the idea I could tell where the bodies were hidden in his garden. There is some science to this idea. I think his secrets remain safe, but I could identify that almost 1% of the DNA sequences from his samples were assigned to Propionibacterium acnes. As a typical skin commensal of adolescents and adults this may well reflect the unorthodox sample container Justin’s compost arrived in…
One of the nice things about helping to shine a light on a poorly-lit area of microbiology is that you can hop between the inane – zits and such – and the serious. I could also flag the presence of Streptomycetes in both samples. Hardly unexpected but worth sharing the knowledge that these are powerful sources of many of our antibiotics. We desperately need more antibiotics, and it was great to wrap up by bringing the great plate count anomaly and the potential of metagenomics for antibiotic discovery to an audience of rush hour travellers across the UK.
This may well be the first time anyone’s tried to broadcast DNA sequencing live. It’s a niche pursuit, but one with strong outreach potential. If you’re thinking of trying this, or a broadcaster approaches you, here are some things to consider.
Ensure you are very familiar with the technology and protocols you will use. Either stick to what you use in other contexts, or trial extensively. Don’t change too many variables in one go. With the recent update to the highly attractive MinKNOW2 we opted to roll back to MinKNOW 1 late last night to stay comfortable.
Don’t lose control of the overall science. Be explicit about what you can and can’t do for technical, ethical or taste reasons. You will need to identify a decent narrative, and that’s easier to do if what you did isn’t the distant cousin of a meaningful analysis.
Conversely, recognize that what you are about to do is science communication, not a research protocol. This may mean compromises, oversimplifications and behaviour generally not conducive to Reviewer 3. I once had great advice from colleague Professor Alun “Mr Frozen Planet” Hubbard in the context of Greenland filming. Put aside any notion you are doing anything scientifically worthy when doing stuff with the media. Anything you do manage (e.g. in camp metagenomics while filming ICE ALIVE) is a bonus.
Try to take with good grace any scientist who then thinks they can play Reviewer 3 in your timeline in response to a technical or scientific choice you’ve made. Enlightened colleagues know better than to do this.
Simple is good. David Eccles rightly points out that at the simplest, strings of ACGT are neat. As scientists we may obssess about N50s this and that. Nobody else cares. Show us the story. The zits not the fastqs.
Contingency is great. We tried to have a contingency for all steps and two contingencies for critical steps. I had two PhD students with prior experience of performance sequencing as helpers. This meant we could parallelize our task and provide mutual support. We did a complete run-through from sample to insight last night, in slow time. It meant a late night for Aliyah and Andre and an all nighter for me, but it gave us confidence we could deliver in real time. We could also switch to Blue Peter mode if everything failed, thus achieving the primary goal of communicating some science about portable soil metagenomics at the expense of technical authenticity.
Doing a slow-time run through flagged an issue with RAD004 rapid library prep and our first sample, and so we kept with the one-pot ligation protocols with ridiculously short incubations today to maintain good strand/pore ratios. On the day the only major changes were to the times of the interviews.
16th of May, afternoon
Get garden soil samples (collected from our VC and from Justin Webb)
Set up working area as an exhibition stall in the recording venue (our arts centre)
16th of May, evening into night – Arwyn’s lab
Extract metagenomic DNA from 6×0.25 grams of each soil sample, quantify and pool as needed. This should give us buckets of DNA to play with.
Prepare Josh-style one pot ligation libraries in duplicate for each sample (1h), run one per MinION. Stop at a minimum of 50,000 reads per sample.
Convert to fastq.gz and upload to http://kaiju.binf.ku.dk/server to generate metagenomic profile (estimated time: 30 minutes)
Prepare a briefing document for Justin on interesting species/stories
17th of May- BROADCAST DAY -Arrive at Arts Centre at 0500h for 0600h show start.
050X:Helpers (Aliyah Debbonaire, Andre Soares) to do rapid soil extraction on 1 sample each (ET: 35-40 minutes)
Note: BBC happy for ambient noise, but TerraLyzer is ridiculously loud. Do bead-beating outside.
Arwyn to prime flow cells, qubit, help etc.
Backup: if DNA extraction poor, use pre-extracted material.
0650: Arwyn on air with Justin. Explain about microbiomes and portable sequencing.
0700: RAD004 library prep and load flow cells.
Backup: use aliquot of LSK library.
Backup2: we will have the lab processed data.
0800: stop, use live called fastq for kaiju (ET: 25-30 minutes).
Backup: Andre will have Kraken on site in event of network trouble.
Backup 2: We will have the lab processed data.
0840: stop, prepare for return to Justin.
0850: Reveal on air. Pick out a few species of interest.
0900: Show off air.
0901: Chill out and clear up.
Your schedule for achieving technical steps will not figure heavily in the priorities of a producer. Plan accordingly. Provide a contingency step-in for when someone has to be interviewed, for example. Recognize hurry-up-and-wait is inevitable.
Finally, enjoy it. This is not a life-or-death application of the technology, so relax.
Thanks to: Justin Webb and BBC Radio 4 for making this really easy for us, Oxford Nanopore Technologies for kindly providing the key consumables needed in very short order, Aberystwyth University’s comms team for NASA-like can-do and efficiency and in particular Aliyah Debbonaire and Andre Soares for maintaining the best of humour while working very long hours. I’m a lucky PhD supervisor.