Our hearty thanks to all those who are live-tweeting from AGBT (and a sad shake of the head to those presenters who aren’t allowing tweets). We’ll use this space to keep track of the blogs and tweet compilations from the conference, as well as commentary on any big news.
Today perhaps the most interesting (if not earthshattering) news was from Illumina. They provided a little more detail on the HiSeq X Ten and the NextSeq 500. With the NextSeq they revealed that is uses isothermal sequencing, something that we hinted at earlier this week. We’re still a little confused about the “2 dyes, no, it’s really 4 dyes but just two channels” issue. We’ve asked for clarification and will report back if we learn anything. They also announced a new NGS startup accelerator program – kind of an interesting idea to juice your R&D program.
Today was a bit more difficult for attendees as there were a number of concurrent sessions – some pretty hard choices had to be made. Again, there weren’t any big surprise announcements, but still lots of interesting talks. There was quite a bit of discussion about the tweetability of talks (see blog list below) and a bit of handwringing over the new reference genome GRCh38 and all the reanalysis it means (it’s a bit like bad-tasting medicine – you know it’s good for you, but you don’t want to take it). And Stephen Fodor, Affy’s founder, finally revealed his new ‘digital quantitation’ company, Cellular Research. It looks like there are lots of great talks lined up for Friday (again, with some hard choices), but we’re especially looking forward to the tweets from David Jaffe’s talk on nanopore data (presumably Oxford Nanopore’s), Barrett Bready’s update on Nabsys, and Hesaam Esfandyarpour’s talk on Genapsys. (We hope the talk is more interesting than the title: “The GENIUS™ Platform: A Next Generation Sequencing Platform That Exceeds Quality and Cost Goals for Universal Deployment In and Out of Core Laboratory Environments”)
The early favorite on Day 3 appeared to be Joe DeRisi who by all accounts gave a riveting talk on the scramble to use NGS to help diagnose a case of acute encephalitis. He even managed to pull off a daring live demo (which is always asking for trouble) of uBiome’s taxonomy browser.
But the clear “buzz creator” of the day was David Jaffe of the Broad Institute with his talk on “Assembly of Bacterial Genomes Using Long Nanopore Reads”. First, and most importantly, the talk included real Oxford Nanopore data – there were rumors that it was only going to be simulated data. That said, Jaffe’s group didn’t produce the data – it came from MinIONs run by ONT. And, in what was surely not coincident timing, ONT sent out email invitations to their MinION early access program (MAP). Here’s what we learned from the talk.
Mostly what we’ve heard before – there are 512 nanopores/cm2 (on the MinION) and each pore can read sequence at a rate of 1-100 bps (the presented dataset was run at 25 bps). The new thing was the read profile of overlapping 6mers -presumably meaning that 6 bases at a time occupy the nanopore.
In the presented dataset the average read length with 5kb, with some reads up to 20kb. Those are nice long reads, but not that impressive in the face of what PacBio is doing these days. It was stated that the read length was limited by the quality of the sample and that longer reads will be possible with better quality samples or more careful prep. This is in line with what we’ve heard from PacBio – prep is now the limiting factor for read length. (However, it should be noted that Nabsys seems to not have a problem getting >100kb ‘reads’.)
The read quality seems to have an unusual profile. There are stretches of perfect reads (84% of >5kb reads have at least a perfect 50mer and 100% have at least a perfect 25mer) that are interrupted by ‘error-blocks’ that appear to be systematic artifacts. These error-blocks are dominated by indels, mostly deletions. ONT is working on improving this through the use of multiple pore types (which presumably will have non-overlapping error profiles).
Overall the response to the talk seemed to be a mix of excitement and disappointment. It’s real nanopore data and it has some interesting potential uses, but it’s not yet living up to what they promised two years ago. The good news is that people started receiving their acceptance letters to the MinION early access program. Getting these sequencers into the hands of researchers can only be a good thing. Can’t wait to see the interesting experiments people come up with!
The other highly anticipated talk, at least by us, was that of Hesaam Esfandyarpour from Genapsys. Unfortunately, it didn’t have quite the same impact. First was the ‘big reveal’ – a sequencer that fits in a lunchbox! “Fits in a lunchbox” isn’t catchy enough, so GenapSys is going with “A footprint the size of an iPad”. Either way, it’s pretty small. The ‘big reveal’ was then followed by… well, not much, apparently. There was no data, instrument pricing wasn’t mentioned (but it should be pretty cheap) and the official launch is at least a year out. Esfandyarpour did talk about the consumables – three chips with outputs of 1Gb, 10Gb and 100Gb. But the pricing was a little confusing as all three seem to be ~$300. Perhaps they’re following Ion Torrent’s strategy of releasing new and improved chips over time while keeping the price constant. The only other bit of news was the immediate availability of the “GENIUS Club”, an early access program for their sequencing platform (the GENIUS system). A $2500 refundable deposit buys you 15 runs on a pre-launch system. The response to the talk seemed to be a mixture of disappointment and frustration at having wasted 20 minutes (and missing other talks as this was during one of the concurrent sessions).
Another day, another series of impressive talks, but still no ‘big news’. People seemed especially appreciative of Jeffery Schloss’ presentation on the history of the NHGRI’s efforts to enable the $1000 genome. Another talk that we would have liked to be at was Daniel MacArthur’s talk on analyzing 50,000 human exomes. (Although, as one tweeter lamented, we don’t envy his task of reanalyzing them with the updated reference genome GRCh38!)
Ion Torrent gave some updates at their workshop. Hi-Q, their improved PGM chemistry, seems to be making progress, with customers seeing ~65% reduction in homopolymer errors. Ion Chef, an instrument that handles template prep and chip loading, is still in early access, but it’s showing promising results. The long awaited Proton PII chip is still in the testing phase. The results presented were reasonably promising, but they’re not even doing full chip runs (presumably due to issues with the necessary 5Gbps transfer speed off the chip). Early access has been pushed back to May/June and the full launch is now the second half of 2014. That’s a long time for Illumina’s recently launched NextSeq 500 to gather some steam.
Overall it looked like another great meeting with lots of useful (and sometimes entertaining) presentations. It gave us our first (if somewhat disappointing) glimpse at Oxford Nanopore’s data, but in general there were no big surprises. Simply by the lack of any substantial new competition, Illumina came out as the big winner, with PacBio a close second with their continually improving long reads. And once again, we give our sincere thanks to all those who tweeted and blogged from AGBT. The experiences and thoughts you shared with the rest of the scientific community who couldn’t make the meeting were incredibly valuable.
- Illumina’s preview of their AGBT activities
- Illumina – Genomic Software Solutions get Spotlight at AGBT
- Illumina’s Day 1 summary
- Illumina User Group Meeting
- Video from Fluidigm
- PacBio Day 1 and 2 update
- Lex Nederbragt defending the ‘no tweet’ policy.
- James Hadfield’s plea for ‘tweet friendly’ talks.
- Keith Robison discusses the impact of the new GRCh38 human reference.
- Tracking discussion of the Jaffe/ONT talk by @homolog_us
- Lex Nederbragt’s take on the Jaffe/ONT talk
- Jaffe/ONT talk review from @nextgenseek
- In Sequence story on the Jaffe/ONT talk
- Nature story on the Jaffe/ONT talk
- Genetic Engineering News “AGBT Report: Sample Prep Solutions”
- Dale Yuzuki’s thoughts on Jeffrey Schloss’ talk
- NGS Perspectives’ summary of AGBT (with links to some interesting video interviews)
- Dale Yuzuki’s review of David Jaffe’s ONT talk
- Lex Nederbragt’s review of AGBT
- Review by Decibio
- PacBio Day 3 review
AGBT Twitter Streams:
- AGBT tweet cloud from the Yaniv Erlich
- A ‘no ads’ tweet stream from Brian Krueger
- A ‘commercial-free’ tweet search provided by…?
- AllSeq’s tweet archive (“Now with extra ads and commercials!”) – #ABGT14, #AGBT2014, #AGBT (but none of that ‘trance’ stuff)
- Twitter ‘news’ from SeqComplete
- Auto-generated tweet archive by Stephen Turner
- @nextgenseek captured the @nanopore inspired twitter storm
- Auto-generated summary from Eventifier
- Bekah McLaughlin’s storify of Thomas Gilbert’s talk
- Tweet stats via Robert West