Agilent apparently saved any mention of their LaserGen acquisition for the Q&A session (for the main presentation, slides & transcript). It's in the indefinite future -- plans to launch it in 2020 have been shelved -- and they don't plan to build a general purpose sequencer but focus on the clinical market (see the 360Dx article linked below).
"On the genomics front, company executives responded to an investor question about the status of Agilent's internal clinical sequencing development, which the company took on with the acquisition of #Lasergen ...— Albert Vilella (@AlbertVilella) January 15, 2020
So, they're going to deliver a fluorescence-based short read technology long after they acquired it, long after many players have staked their turf and without pitching into the research market that generates applications and user base. Sound familiar? It's QIAGEN's strategy, which they were finally smart enough to bail on this year. Agilent thinks they are smarter and can make it work -- even after a flood of new entrants are likely. We already have Genapsys in the market. Omniome is champing at the bit. Multiple other potential new entrants likely to launch in the next few years. About the only weakly positive thing I can say is that a lot of Illumina's patents will be expired by the time Agilent launches so they won't have to worry about that. More on that front later this month if I get my rear in gear.
ThermoFisherThermoFisher announced a deal with LabCorp to use Ion Torrent technology for LabCorp's oncology business, but otherwise nada on the sequencing front. ThermoFisher is often rumored to be portraying acquisitions in genomics, previously thought to have made a pass at QIAGEN and thought maybe interested in PacBio. Hard to shake the feeling that ThermoFisher is a retirement pasture for sequencing tech, where they live out their lives with a few sweet little deals like the LabCorp, but their glory of being in the big race is long over. (transcript)
10X GenomicsNow onto some actual genomics companies. Somewhere in the last seven years 10X Genomics has grown to a market cap of $8B, which is quite impressive. (audio, slides).
10X on the Chromium side is very focused on the various single cell applications. Further growth in the capabilities here are promised; whereas when Chromium launched profiling 10K cells in an experiment was considered impressive, a slide suggested 100K for this year, 1M next year and on to 10M sometime in the future.
I realize this is a financial conference -- ol' John Pierpont didn't care about science -- but Saxonov did make some strange comparisons. He noted that there are 50K real-time PCR instruments in the world and suggested that this represents the ultimate market scale of Chromium -- and quite directly suggesting that Chromium will replace those instruments. This of course is absurd; many of those instruments are being used very differently than a Chromium; I love qPCR for assaying a few things on many independent samples and doing so very rapidly. Plus the cost of a Chromium experiment is radically greater than qPCR; even if 10% of those labs could afford to buy the box they'd probably not have the money to run many assays.
A bunch of time was spent on their Visium spatial profiling tool built around the technology acquired with Spatial Transcriptomics. With DNA-linked antibodies, one can simultaneously profile both RNA and proteins on thin tissue slices, and they even showed results with archival FFPE samples. Takes me back to my Infinity days and to me this sort of capability is catnip. I'm resolving to more closely follow this emerging field (see also NanoString below). 10X promised it will have a big presence at AGBT.
One curious bit on two slides -- one for Chromium and one for Visium -- is a workflow schematic showing how 10X has all parts of the pipeline covered and generating revenue except the sequencer piece.
I like he parallel construction of the slides, but the prominent box calling attention to only 25% of customers owning their own sequencing box is curious. Saxonov didn't talk much on this point, so we're left guessing. Is 10X considering acquiring sequencing technology? It would seem a bit odd -- playing the field has many advantages. Also no mention of long read technology -- there's a few but exciting preprints on using 10X in conjunction with nanopore to generate full-length cDNA from single cell preps. Just for buzz I'd recommend 10X test how well nanopore can read their current libraries intended for Illumina coming off the various preps -- immunoprofiling, single cell RNA, spatial RNA. Particularly with ONT pushing accuracy and tuning up their issues with short fragments, there might be a great opportunity here.
CEO Serge Saxonov
nanoStringGotta get one thing off my chest up front: nanoString technology can generate beautiful gee-whiz images using spectacular single molecule approaches. So why is the background for their title slide a radioactive Sanger gel???
On the corporate front, nanoString detailed their recent partnership with Veracyte on cancer diagnostics (in particular the Prosigna breast cancer prognostic) which refilled their bank account and also led nanoString to drop some internal development efforts in that area.
NanoString is also going big on spatial profiling, again with dual threat of RNA and DNA-tagged antibodies on the same sample. They launched last year and see great growth. Interestingly, the speaker shouted out to 10X Genomics at least twice as an exemplar of success in the single cell market - a Chromium is even depicted on two slides; 10X will probably be their biggest competitor. Certainly a well-funded one; NanoString's market cap is about 1/8th 10X's and so their access to capital likely proportionately more difficult.
For their GeoMx (which I keep wanting to correct in my head with the 'N' every time I hear it -- ironic for someone who has spent a lifetime trying to dodge inserted N's in their surname), nanoString emphasized growth in capacity and the imminent launch of sequencing-based readout of the GeoMx assays. So 84 genes in panel readable with nCounter at launch, 1800+ promised for this year using sequencing and 18K by sequencing in 2021 if all goes well
Hyb & Seq is now targeted for launch in 2022. For some clinical applications, nanoString believes that blood draw to results time could be from 3 to 5 hours with less than half an hour of hands on time
nanoString is hosting a big pre-session at AGBT -- "2nd Annual Spatial Genomics Summit" on Sunday of that week and will have 2 talks and 8 posters in the main session.
PacBioPacBio didn't present at JP Morgan -- the acquisition by Illumina being called off probably squelched any opportunity to get on the agenda -- but the Plant and Animals Genome conference was going on at the same time and that's a major user community for the company. PacBio's Michelle Viera was kind enough to share her slides with me.
No radical announcements, but a number of evolutionary points. PacBio plans to bring Circular Consensus Sequence processing onto the instrument; it is currently run off-instrument. CCS computation continues to be shaved of compute needs. This will simplify workflows. They also are aiming to generate a complete quality human genome from a single Sequel II 8M flowcell. A twitter conversation on this point suggested that this might require only about a 2X boost in performance, if that genome is only around 20X coverage in HiFi consensus reads -- which a body of work suggests is sufficient. Indeed, in one slide 18X coverage was used for high accuracy, high precision variant calling on a human sample. HiFi-based assembly from 48Gb of CCS reads took about 31 hours.
On the sample prep side the company is emphasizing there is a single library prep for all workflows and that DNA inputs can be as low as 100 nanograms -- except for a new ultralow kit that uses amplification to sequence from only 5ng. This was used from a single sandfly to generate 35.9 Gbp of HiFi data with a mean read length of 10.8Kb and mean quality of Q36. The resulting assembly had a contig N50 of about a megabase, with 363 megabases split into 977 contigs and 97% BUSCO completeness.
For human variant calling, the same sample gave very similar results with 18X coverage generated either with Standard HiFi or Ultra-low HiFi, though for indels both recall and precision dropped by many percentage points.
HiFi is also touted for metagenomics projects -- the high quality reads enable direct gene prediciton with assembly.
Iso-Seq is the other premier application on Sequel II. A new Iso-Seq express kit requires only 1 day, only 300 nanograms input, has no required size selection and is claimed to reduce the cost by a third. An Iso-Phase tool enables detecting transcribed polymorphisms. PacBio estimates that $1300 can get you a full transcriptome for a mammalian tissue with all isoforms, or go for lower coverage and multiplex samples to get a reference transcriptome for $180 a sample. UMIs and single cell protocols are planned for the near future.
One other note: I love the bullet saying that in the future the software tools will support demultiplexing more than 1000 samples in a run. Let's hope the barcode supplies are built to match that.
Oxford NanoporeONT also had a big presence at PAG and officially announced that they are taking orders for R10.3 flowcells -- about a month later than the mid-December launch trumpeted at the Nanopore Community Meeting. A reminder I (or I'm happy to let someone else do it!) should track the slip in ONT launch dates.
PacBio -- "human genome on single flowcell" -- 2X improvement needed? HiFi analysis going on box