So you need some DNA sequencing? (pt 3 conclusion.)

I've stretched this topic out farther than intended, so I'll conclude directly:

-from looking at the map, there seems to be 4 types of sequencing centers, each with different strategies and hardware needs:

1) medium-large installations (Broad, BGI)
2) fee for service centers
3) genomic (academic) centers with a commitment to genomic research (5-10 sequencers.)
4) academic centers with a small exposure to genomics (1 or 2 sequencers.)

Each of these will have different rates of adoption of NGS technologies. Here's how I'd characterize each of these centers future behaviors:

#1) medium-large centers: all about throughput and cost, with less regard for specialized instruments or needs, these centers also already have a substantial investment in hardware and informatics, so the winning hardware providers will be the ones that plug in best into the existing hardware and informatics. It will be a whole lot easier to integrate the latest generation of Illumina technology than to pivot 90 degrees to integrate a novel technology.

I expect that the number of medium-to-large centers rises, as the cost/sequencer falls and the start-up cost of a new sequencing center falls. I don't know if research demand for such centers is here yet, but I think several institutions will launch ~$10M fundraising efforts for a new genomic research center, as much for their economic development/headline value as their scientific value. (Example: the former Ignite Institute, which landed at Fox Chase.)

2) fee for service centers: I selected the first 5 US service providers that I could think of (Asuragen, Beckman Coulter, Cofactor, Expression Analysis, Seqwright), and was surprised to see their total capacity was 26 sequencers among them. The absolute number could be outdated or inaccurate for a number of reasons, but the point is that the service centers aren't big consumers of technology. (I'd guess, though, that they run at higher capacity utilization than most academic sequencers.)

The fee-for-service centers also tend to have more than one technology platform in-house.  As demand grows, the fee-for-service centers will add capacity in a nimble, savvy, but serial fashion, spread among whichever technologies are requested by their customers, and which provider has the best performance/value proposition at any given time.

3) academic genomic research centers. much of the research at the genomic centers will be tied to clinical trials, so this group will be very sensitive to FDA approval of a sequencing device, and not very sensitive to throughput/performance though turnaround time may matter if the clinical trials are looking for the sequencing data to guide treatment. I'd expect this group to hang with the Illumina technology for the foreseeable future. They're the most likely platform to receive FDA approval. (Unfortunately, this isn't likely to happen soon, if the FDA approval of microarray platforms is any indication. As a forerunner to sequencing, Affy got their microarray platform approved by the FDA (in 2009?) for clinical diagnostic use, but I've heard that it wasn't easy, and the approval is not too broad.)

#4) small-time centers: the largest market in number but smallest in $$$. This market won't grow significantly until clinical adoption of DNA sequencing becomes widespread, and even then the biggest customer may be the pathology labs, not the bench researchers. In this case, I'd expect this category to largely adopt either the nanopore or Ion Torrent technology, as much for simplicity as for throughput and cost.


After this analysis, I am surprised that the opportunities for new platforms such as Oxford Nanopore are not as obvious. The newcomers may still be a success, but I think we're still a few years away from the inflection point in the growth of sequencing hardware.

So you need some DNA sequencing? (pt 2.)

Following up Sunday's post based on the most-excellent Next Generation Genomics: World Map of High-throughput Sequencers, let's try the same experiment on an international basis.

(But first, a caveat: after spending more time with the map, it is pretty clear that the census is really of non-profit academic institutions & service providers. There's really no incentive for a for-profit R&D to report their sequencing capacity.That said, shouldn't commercial DNA sequencing activity track academic efforts, or are some countries biased towards (or against) for-profit R&D?)

DNA sequencers by country:

USA: 827
China: 216
UK: 140
Germany: 112
Canada: 73
Spain: 56
France: 36 (and very well dispersed geographically)
Japan: 35 (there's much consternation that Japan is falling behind in sequencing.)

Italy: 31

South Korea: 29

Taiwan: 28

Switzerland: 27

Sweden: 25 (rest of Scandinavia (DEN, SWE, FIN, NOR: 30))

India: 23

Singapore: 16
all of South America: 18
all of Africa: 12 


Undeniably in LAST place: Indonesia. Human population: 237M (4th most populous country) DNA sequencer population: 0

Observations:

-Even I'm surprised at how dominant the USA is in sequencing. A quick glance suggests that half of all sequencers are located in the USA. As a policy point, this is worth exploring in another post of its' own (US healthcare model vs. socialized medicine, anyone?), but for now, let's just debate the drug discovery implications. Is there any reason to believe that at least half of the RX & DX innovations won't come from the USA? And if so, is there any reason to believe that countries with socialized health systems aren't free riding on innovation funded by the US system? (Oops. Sorry. That last question just slipped out.)

-As nice as it is to have a large lead in hardware, keep in mind this also means that the USA will led the world in stranded assets and hardware depreciation. Just to illustrate, if the average American sequencer represents a $300k investment, there's a quarter billion dollars of value depreciating quickly.

-wouldn't it be great if LIFE or ILMN donated a few sequencers - even used/last generation - and a supply of reagents to a few universities in Africa? Maybe LIFE could use the US tax code to sell more Ion Torrent machines to existing sequencing centers by facilitating tax write-offs of SOLID equipment donated to universities in Africa or South America? (Company X donates $150k in SOLID equipment to the University of Witwatersrand (Nobel Laureate Sydney Brenner's alma mater). Company X then uses their $50,000 in tax savings to buy a new PGM.) (Does such a program already exist?)

-I'm kind of surprised at the modest presence of the Gulf States. 4 sequencers in Qatar, zero in Bahrain/Dubai/the Emirates.

-I stated that the map listed BGI as having only 15 sequencers in my Sunday post. Turns out that the bulk of BGI's sequencers (166) are in Shenzhen. Apparently even the Beijingers outsource to lower-cost parts of China.



Next up: a look at what this all means.

The M&A game….

…..featuring Illumina + Roche again and an addendum to my post last week about the Affymetrix-eBioscience non-deal. What do these 2 deals have in common?

Luke Timmerman @ Xconomy has a good piece this week with 5 reasons why the Roche + Illumina deal isn't right for Illumina. I shared reason #6 in the comments:

• "…..because the touted “total solution” provided by a Roche + Illumina combination is a fairy tale. Illumina sells equipment. Roche sells drugs and diagnostics. What tiny bit of equipment that Roche sells (454) hasn’t done well. I don’t see how selling Illumina equipment makes Roche’s drug or diagnostics businesses any better. What “total solution” becomes enabled by the combo that isn’t possible by Roche just buying a roomful of Illumina (or someone else’s) sequencers?" 

What the ILMN-Roche and AFFX-eBiosciences deal have in common is that both deals are now an exercise in game theory. Consider ILMN's options:

1. Accept Roche's bid. (notgonnahappen. Roche's offer is ~$6 below the current price) 
2. Adjust the terms: wrestle for a higher bid from Roche or find another bidder to up the price. 
3. No deal. Win a proxy fight by making the stand-alone scenario more real and financially attractive. 

Likewise, consider eBioscience's options:

1. Accept AFFX's likely revised downward terms, though still rich, in order to allow AFFX to win debt financing of the acquisition. 
2. Adjust the terms by selling to another suitor, likely at a lower price than AFFX's rich offer. 
3. No deal. No liquidity for investors. 

Timmerman argues Illumina shareholders should vote to remain independent for largely qualitative reasons. Unfortunately, I think the decision to be made by shareholders is much more cold and quantitative: what's the better risk-adjusted net present value?

1. Roche's $44.50/share bid, (again, notgonnahappen.) 
2. a sweetened bid, or 
3. the capital gains in future years from selling ILMN shares after the company stock re-appreciates. 

Putting some #'s to #3. Using round figures, ILMN is at $50/share, and had a previous high of $80/share. Holding ILMN stock for 2 years to see $30 in appreciation would require an annual return to equity holders of 26.5% - a not unreasonable scenario, particularly in such a growing industry. The problem is, the $30 gain offered in the future (over two years) can be made a lot less relevant with a sweetened bid 'now' by Roche.

What if Roche offered an additional $2B, which would raise the ILMN offer to $60/share, or about a third of the 2-year gain upfront? This might be hard for ILMN shareholders to turn down, especially if the offer is cash-heavy.

To me, and likely to both ILMN shareholders and Roche management, the outcome is determined largely by your appraisal of ILMN's NGS technology. If you think ILMN is in danger of being passed by Ion Torrent or Oxford Nanopore, you take a sweetened offer from Roche. If you think ILMN has the tech to stay on top, you probably hold your shares (or, if Roche, increase your bid.)

All of this says to me that we should be on the watch for a public unveiling of ILMN's future NGS tech, or their roadmap as such. (Via a press conference or an analyst day, or the like.) ILMN is currently touting NGS prices of ~$5,000 per genome. If they can demonstrate a technology (or path) that drives this number down into ONP's ballpark (~$1,000), expect ILMN to stay independent. If not, ILMN will take Roche's best offer.

Roche has already played their role in this game, as they played the "you know you're not the only fish in the sea" card - even though the whole world knows that there isn't an equivalent alternative NGS investment available. (Unless you think PACB or GNOM make for good back-up plans.) I interpret this as Roche saying that they're open to paying a bit more for ILMN - otherwise, they'd play either the "take it or leave it" card.

Nearly six weeks have passed since Roche's hostile bid and yet Illumina hasn't shown off any reason for shareholders to expect ILMN stock to pop as an independent company. Be on the lookout for either a sweetened Roche bid or a big ILMN tech exposé.


While ILMN is looking for paths to increased valuation, eBioscience must be looking for how to avoid too much decrease in valuation. It looks like AFFX can't do the current $330M deal, as lenders are pulling their financing. They could seek another bidder, but presumably they held an auction before accepting AFFX's bid, and know the possible range of offers. At 4.7X trailing revenue, the AFFX offer is very rich.

As a mostly-commodity provider, eBioscience probably still wants to get a deal done, even if AFFX can't honor the proposed terms. (btw: there are differing reports on whether AFFX's offer is all-cash or 50/50 cash/equity.) Would eBioscience rather take a tweaked deal from AFFX at say 90% of the value, or - as they are a growing company - sell a year or two later to someone else at a reduced multiple? (say $80m in 2012 revenue x a 3.75X multiple (=$300M.)) Chances are, this offer from AFFX represents the best and most lucrative chance for liquidity for eBioscience shareholders that they are going to see for a while.

The best outcome here for eBiosciences is to negotiate a sale at a point between their best alternative purchasers' price and the $330M, or to alter some deal terms to slightly reduce the value of consideration from AFFX. eBioscience could keep the same headline number, but accept a mix heavier on equity than cash, for example. Or, eBioscience shareholders could provide the debt financing themselves, in the form of an earn out or milestone payment from AFFX.

Unlike ILMN, I don't think that eBiosciences has to worry that their suitor will have a change of heart. If the financing gap can be bridged, the deal will happen. At this point, it seems to be a matter of how much less lucrative terms eBioscience is willing to accept and whether this figure works for AFFX's bankers.

So you need some DNA sequencing? (pt 1.)

There's an absolutely brilliant map and website devoted to taking a census of DNA sequencers around the world, and through it, you can monitor the tug of war between the placement of instruments in large DNA sequencing centers, and in individual labs (or core labs) on a onesie-twosie basis.

(The map may be a little dated - it shows only 15 sequencers @ BGI, for example - but the brilliance of the map is in how the data was generated. Do a little web surfing for the WWII spy technique that spawned the map.)

Anyway, a VERY interesting story is told when looking at the USA map (select USA under the country pull-down. Unfortunately, I can't link to this specific page.) The sequencer census really indicates who and what are on the edge of the genomic revolution.

A rough read of who's ahead by sequencer placement:

Boston: 129 (Broad Institute = 104)
St. Louis: 85 (includes 11 at Monsanto)
San Francisco Bay Area: 58
Washington, DC area: 51 (with an additional 23 in nearby Baltimore and Frederick, MD.)
NYC area: 47 (includes 18 @ CSHL on Long Island.)
RTP, NC: 37
Houston: 35 (driven by MD Anderson.)
Toronto: 33
Southern California (LA + SD): 27
New Haven, CT: 23 (birthplace of 454 and IonTorrent.)
Philadelphia: 22
Albuquerque, NM: 19 (Sandia)
Montreal: 18
Memphis: 12
Seattle: 5

most other US metropolitan areas and universities had 1-5 DNA sequencers listed.

(Note: some of the math above is fuzzy, as figures change slightly depending on how far you drill down on the map. Plus, the math is skewed by self-reporting and non-reporting. For example, there is a paucity of hardware listed in 'PharmCountry' (NJ & eastern PA) and there's virtually no privately owned sequencers in the Bay Area, but you can be sure that both Big Pharma and Genentech have some sequencing hardware on campus.

What does this all mean:

-you can see which cities are in position to lead the genomic revolution, and which are likely to be laggards. For example, according to the map, there is only 2 DNA sequencers in the city of Chicago. There's roughly 25X more genomic activity on the i-270 corridor outside DC.

• I was surprised by the strong figures in St. Louis and RTP. Both locations have strong ag-bio efforts, so you might be able to extrapolate that the earliest exciting NGS uses will come from ag-bio, and not cancer genomics.
• Anyone else as surprised as I am that NYC was in the top 5? Certainly there is great science at places like Rockefeller, MSKCC, and such, but I wouldn't have guessed top 5. With some announced initiatives, NYC should stay near the front.

-you can see funding philosophies in action (if you squint). There is a tremendous amount of hardware in Canada relative to population, representing governmental support skewed towards hard assets rather than funding annual research. (And the good work of bodies like Genome Canada.) However, with the rate of technological innovation in sequencing, hardware represents a quickly depreciating asset. Maybe the better use of the funds was to pay for outsourced sequencing. (Perhaps this was the case in California, where the inverse was present - less hardware than you might expect.)

-if I had to bet which institution will lead in the adoption of DNA sequencing in patients, MD Anderson would be my first bet. My quick survey suggests that they're the practicing center with the most NGS hardware. Yale isn't far behind.

-on the other hand, some NCI comprehensive cancer centers were extremely lame, including my hometown University of Virginia, Northwestern U, and OHSU. 1 sequencer each? Lame!


Tomorrow I'll do some analysis based on the type of hardware, and take a guess at what this means for adoption. I'm especially interested in analyzing the question of what the distribution in sequencing is likely to be between CRO/service providers and sequencing with internal assets.

NGS & DX?

There's an interesting conversation going on about error rates in DNA sequencing in the Genomics (NGS) group on LinkedIn. Some are wondering if development of DNA sequencing diagnostic applications will be delayed by the experienced error rates (up to 4% on some platforms, including Oxford Nanopore.)

My take: I think the barriers to adoption of sequencing technologies as diagnostics are:

-any error-intolerant application is still likely to rely on RT-PCR for a while to come. (Example: detecting specific BCR-ABL mutations in CML patients.)

If you have a specific gene of interest, or even genes (up to about 10 or 20, depending on who you listen to) 

-PCR still wins the day, because of accuracy, speed, cost, privacy concerns, and the fact that PCR apps have familiar payor and FDA tracks. (Many PCR assays code for reimbursement <$300, so NGS still has a way to go to win on price.)

NGS, on the other hand will be used for broad discovery and in cases where patients are willing to pay out of their own pocket at least until the economics change, and the FDA approves a platform/assay combo such as Foundation Medicine. I'd say that we're at least 2 years away from that, regardless of error rate.    

Two other NGS points, neither worth a dedicated post for now: now that Oxford Nanopore and LifeTech are both promising ~$1,000 genome from new tech platforms:

-what does the future hold for BGI (Beijing Genomics Institute) that has made a name for itself by buying roomfuls of largely Illumina sequencers? I'd like to be a fly on the wall when someone suggests that they put 10's of millions of dollars of Illumina equipment out to pasture and invest further millions in new GridIon or Ion Torrent equipment.

-will Roche drop their Illumina takeover bid? A ~$6B hostile takeover of the former leader makes less sense now. It will also be interesting to see if ILMN's board changes their mind, and sells now.