London Calling was two weeks ago, and I still haven't written anything beyond the write-up of Clive Brown's talk (note that many of the talks are now view-able via Oxford Nanopore's website). I did finally make some headway on Storifying the tweets. After several self-inflicted wounds (starting with failing to read my write-up of the previous experience of trying to post to Storify from the command line, but also from failing to record the details of that final critical command), I did succeed. This time I decided to group the tweets into several broad categories -- with tweets potentially showing up in multiple categories, though these are a bit rough-and-ready and probably most folks will want to check multiple or all of the stories. One major value I derive from this exercise is having them to refer back to, so I'll probably feel the pain of erratic characterization the most (though I also have the tweet database, so I could reslice these in new ways). Anyway, the categories are: de novo Assembly, Bioinformatics and R9, Real Time and Read-Until, PromethION, Microbiomes, Clinical/Human (except Infectious Diseases), Infectious Disease and Biowarfare, Field Uses plus finally Library and Sample Prep (another one that failed initial upload; includes Zumbador and VolTRAX tweets).
There's supposed to be a final one encompassing everything that was left over, particularly general tweets about the conference. Unfortunately, Storify balked at that one -- have to look again and perhaps break the set into at least two pieces. (nope: worked with one)
If I was going to make a categorical statement, most of the talks at London Calling this year were leveraging MinION's small size, portability or speed-to-result, with many utilizing combinations of these three. There were a few talks about general genome assembly and some others outside this scope, but by far this was dominant. Most of these applications also benefited from (or required) long reads. I think there were at least 4 talks on HLA typing (needing long reads), multiple on infectious diseases (both viral and bacterial) and several on working outside conventional laboratories. Looking at what Clive covered, a lot of the tech promises from ONT will be a boon for these spaces. I'm particularly hung up on the rapid 1D kit, anxious to try it both at work and at home. Zumbador would clearly be useful in the field, eliminating all the accessory hardware for DNA extraction and library preparation. Well, at least if as you are doing genomic preps from easily lysed samples, though portable PCR devices are sprouting like weeds -- a quick Google search found Palm PCR, miniPCR, R.A.P.I.D. BioDetection System, Slim PCR Cycler, Sentinel and more. VolTRAX looks like it could be portable as well, and while PCR won't initially be supported ONT has mentioned it as a goal.
ONT has already delivered local basecalling to eliminate the need for a cellular connection, though apparently the ZIBRA team in Brazil is finding that the local basecaller on a 4 core laptop falls far behind the MinION's output rate (which gives pause for the SmidgION concept in current form; could a smartphone actually handle the basecalling, let alone calling plus analysis?. SmidgION looks great on a slide, but a number of persons (including myself) are skeptical that you really want to power off the phone -- after the phone battery is used up, what then?
This is as good a place as any to talk about the stunt sequencing that is really tearing at my loyalties: MinIONs in space. This summer a MinION will travel to the International Space Station and be used to sequence a library prepared on the ground. An earlier test of running the device on "Vomit Comet" parabolic flights yielded a humorous video but as an absurd travesty. Given that these flights give brief bursts of weightlessness, followed by bursts of very high G-forces, the exercise was of dubious relevance to operating the devices in space. The real test is whether the equipment survives the high G-forces and vibration load of launch.
I don't have much doubt the MinION will work in zero-G, as at the scales being operated at fluid properties such as surface tension and attraction to walls of the device will almost certainly overwhelm gravity. In any case, library preparation in space won't occur until some later flight. Finally, beyond the high publicity factor, NASA is a terrible partner for a fast-moving technology, particularly in the manned spaceflight program. After the horror of Apollo 1, the requirements for testing hardware for flammability and such (probably outgassing also) were made strict, which means every new hardware revision will be slow to make it up to space. Finally, it is very, very hard to think of a good rationalization for sequencing aboard the ISS, rather than de-orbiting samples for running on the ground. Ultimately it would be cool to have a sequencer look for life on Mars, Europa, Io, Titan or other extraterrestrial candidates, but that is a totally different set of challenges, starting with a very long, very cold trip to the destination. Undoubtedly a spectacular marketing opportunity for ONT, but yet another dubious stunt from NASA's biology group (I could easily -- and probably should -- run a whole post on what I would see as interesting biology on the ISS).
Okay, rant switch to off and we are go for more prognosticating!
After thinking about the tech announcements, the other dominant trend is to vastly reduce and ultimately all-but-eliminate the technical skill required to operate the platform. Right now, you need to be pretty good at accurate pipetting, running a centrifuge and doing bead cleanups (all at the edge or beyond of my capabilities). The rapid 1D prep is reported to require just three pipet steps, and I think those are all additions. One must transfer product at the end to the chip, which is perhaps one of the trickiest maneuvers due to the need to avoid introducing air bubbles plus some fine motor skills to align the pipet tip with the sample port. But still, that's a huge improvement and definitely brings it into being plausible for me to run. Zumbador is projected to eliminate all the pipet steps, requiring just an alignment of the device over the MinION flowcell. SmidgION, should it actually happen, would utterly eliminate technical requirements, and perhaps require no more hand-eye coordination that a typical kindergartner possesses.
That's all a good thing. Certainly from my point-of-view, those steps are busywork and nuisance, getting in the way of the intellectually interesting part of looking at the data. And certainly it lowers barriers to routine use in molecular biology laboratories. There it is not the skill of pipetting that is an issue, but the learning of a long, multi-step protocol. For education markets, eliminating all those accessories not everyone has (magnet ring stand!) is a plus. This trend will also be important for key field, forensic and medical uses of nanopore sequencing: automation should minimize variation, which is critical.
London Calling does tend, understandably, to being a bit of an Oxford Nanopore bubble. A few speakers mentioned Illumina data (either contrasting the time/cost requirements or using it for hybrid assembly or assembly correction) and a number of talks (particularly on assembly) mentioned Pacific Biosciences. But for the most part, talks are solidly about nanopore sequencing. So it was interesting that just after I got back that the Sequencing, Finishing and the Future (perhaps better known as SFAF) conference convened in New Mexico. I wasn't there, but a number of tweeters were. Actually, I got a bit roped into Storifying that conference -- I had thought about doing it, but then there were tweets from people who thought I had, and I then felt I needed to do it to maintain my reputation! Seriously, curating all the tweets is a great way to absorb and assimilate information from the twitter stream, plus now I can dive back into it.
Anyway, what is so striking is that the only mention of nanopore sequencing in the SFAF tweet stream were remarks on the absence of nanopore sequencing from SFAF. Which is particularly notable since so many of the talks were ploughing similar ground, particularly rapid diagnosis of infectious diseases and their drug resistances. A talk at SFAF using (I think) Illumina was excited to get the diagnosis time for tuberculosis drug resistance down to one week; at London Calling Zamin Iqbal talked about taking one day, two if the resistance signal was weak.
Now, the speakers at London Calling are generally demonstrably familiar with a wide range of sequencing platforms; many run core facilities or have published on multiple platforms. So at LC there aren't blinders. But incidents such as the SFAF talks suggest that Oxford is still a niche platform, with a small (and perhaps slowly growing) band of devotees.
A lot of Oxford's promised improvements could help it break out, but from my particular perspective the big fix is getting the homopolymer problem solved. Because of how the raw signal is parsed into individual events, the R7 basecaller couldn't call a homopolymer longer than 6. R9 chemistry and basecaller are radically different, with the pore strongly interacting with far fewer bases in the DNA, but apparently homopolymers remain a glaring issue. For many of the applications that have emerged, such as identifying microbes and their pathogenic armamentarium, this hasn't been an issue. Similarly, when direct RNA sequencing comes online, for many users identifying the transcript and all the exons in it will be sufficient. But for MinION and PromethION to become major tools for de novo assembly of reference genomes, it needs to solve the homopolymer issue. Some of this may come with better chemistry which is more metronomic in moving the DNA through the pore, and some may come from basecallers that actually try to use the time component of the signals, but it needs to happen.
With summer, the genomics conference schedule goes on a bit of vacation. I may take a crack at Storifying the recent, multiple Pacific Biosciences user group meetings. Festival of Genomics butts up against a computer-free vacation; as a result that may not get any look until mid-July. And as fun as it is to write about reading DNA, I really should write about writing DNA, where big news has happened.