Anyone know details on Church's Polonator Next-Gen Sequencer?

[ECO]

Alright community, it's time to contribute!

I noted a teaser article today at In Sequence (part of GenomeWeb) that suggests that commercial launch might be close behind. Does anyone have any information regarding the commercial launch of this thing? A reference to "Danaher" and a look at their website tells me that they are manufacturing the thing.

There is remarkably little information available on the web regarding the status of the next generation sequencer (often termed the "Polonator") that was born from the George Church lab. I have seen some brief mentions of the homebrew effort in GC's conference presentations, citing a cost of <$150k for a machine capable of very high throughput for very low costs.

So I think this is a good topic to get some input from the community...I know there are a lot of next-gen people out there both reading and writing in the DNA Network. So don't be shy.

[Mr. Gunn]

I don't have the full details, but reading between the lines it sounds like the major improvement is in the software.

[ECO]

Would love to hear what you know! I'm sure there are many interested labs who are pricing SOLiD or Solexa systems...

[pmcget]

We've been in contact with the main players involved in this consortium so I have quite a bit of detail about the machine.
The whole philosophy behind this machine is open source and that may be it's greatest strength and weakness at the same time.

Technology
Compared to the other machines on the market it is definitely closest to the ABI SoLID machine. Basically it is a flowcell/slide on an X, Y, Z stage, with a camera, laser and fluidics and storage chambers for the reagents. It also comes with an application server package(optional). It has a 2 slide setup, so, similar to ABI, it does chemistry on one while it does imaging on the other. One difference is that the slides are split into 18 sections (rather than 8 with the Solid) so altogether you can have up to 36 different samples per run - and that's before you look at multiplexing samples! It uses a sequencing by ligation chemistry (which you can read about in Church's recent papers Shendure et al 2005. Again this makes it similar to ABI solid (I think ABI may have even licensed the chemistry they use from Church). Although with some very important qualifications. The latest information I received the typical operation involved paired end reads where you sequenced 18bp tags from each end. But you don't get a full 18bp read from each end. Instead due to current limitations of the chemistry you get 7bp read one end of the tag 6bp read the other = 13bp. When you read both tags you have a total of 26bps that you can map. So for now this eliminates the possibility of using a fragment library as 13bp won't be enough to map to a mammalian genome.
However the good news for the future is that the machines with the capability to use different chemistries. At one stage I heard that they were trying to license a cleavage chemistry that would then allow them to generate longer contiguous reads - unfortunately I don't know how this is progressing.

There are three big advantages with this machine.
1. Data generated per run - this is currently quoted at 10GBases of mappable data/run and they expect to be able to increase this to as much as 50GBases/run by sometime in 2008. Hence the reason for Church throwing his hat in the ring for the X-Prize. Some of the technical improvements that have led to this can be found in the supporting material to one of Church's recent papers Kim et al. Science 2007 The two big improvements were gained from moving to gel-less slides so that the polonies were in a single focal plane, and by the addition of PEG to induce macromolecular crowding. (This improved the yield of cloned polony beads.)

2. Church is agressively pursuing sequencing cost. Hence he has signed up with the company Enzymatics to produce the reagent kits for the machine. They don't have exact figures yet but they are telling people they will be providing reagents at a price an order of magnitude less than what companies like Illumina and ABI are. As far as I can tell the other companies are probably making more money off the consumables than they are off the machines themselves. Church quoted a reagent cost of around $400 per run at a meeting back in October 2007. This compared to $4000-$10000+ for some of the other machines.

3. The machine cost itself is a fraction of what the other companies are charging: $140-150k compared with $400k and up for Solexa and ABI.

The drawbacks:
1 Training
For most organisations the thing that will probably worry them most is the lack of support and training in library generation. Danaher provides machine training but they won't be able to help with biochemistry problems. I expect that the guys in the Church lab will be far too busy to be able to do the type of handholding that will be required particularly in startup situations. From what we can tell (and all the companies say the same thing) library generation is the trickiest part of the process. It is also the biggest point of failure for experiments with any of the machines. For institutions relying on core facilities to run these machines it's probably not going to be a runner (unless you have a hacker technician/wet lab biologist willing to figure out all the issues with the protocols themselves).

This situation reminds me a lot of the situation with Linux back in the early 90's. There was a lot of resistance to bringing linux into commercial organisations even though it was free - and probably better for the job - than Windows NT Server. But companies need someone to call who will be on site in a specified timeframe to make the machine work come what may. Eventually companies like Redhat and Suse stepped in and filled this market. Maybe another company will take advantage of this opportunity and provide training for the polonator?

2 Chemistry
I think until they can provide contiguous reads of at least 25bp people will be a bit hesitant to go with this machine as it definitely will restrict the types of application e.g. for microRNA sequencing.

Overall
Depending on the projects people want to use the machine for right now the chemistry might put them off. On the other hand if you can use the existing chemistry for your experiment then the cost savings are immense.
Given the technical expertise that is required I can see the leading edge labs embracing these machines and if anything it will probably increase the gap between the scale/quality of experiment they are able to do and what labs further down the foodchain (relying on core facilities) are able to produce.

[ECO]

Awesome post pmcget!

I think I will republish this post on the front page of the site to make sure people see it via the RSS feed.

[JWT]

Web Site

Web Site for Polonator G.007 instrument is now live at: http://www.polonator.org

[ECO]

Web Site for Polonator G.007 instrument is now live at: http://www.polonator.org

Hey JWT...I was going to say...polonator.com confused me!

[Kevin McCarthy]

It's Polonator.org, not Polonator.com

Hi everyone! I am currently attending AGBT, and am with Danaher. I worked very closely with Rich Terry and the rest of the utterly brilliant Church Lab team to create the Polonator. We shipped our first instrument yesterday (ta-dah!!!). It and we are completely open. I am a little busy here at the conference, but am more than happy to answer questions regarding the Polonator.

All the best,

Kevin

[pmcget]

Hi Kevin,
Congratulations on getting the Polonator into production! I think fans of open-source science will be rooting for the success of your machine.

I have a few specific questions that I was hoping you could answer:

1. What are the consumables cost per run?

2. How long does the DNA sample prep take?

3. What is the minimum amount of starting material you can use?

4. What is the current sequencing accuracy of the machine?

5. How many bases of mappable data can the machine generate per day and per run?

Also any hints on what the spec of the A.008 will be?

[Kevin McCarthy]

pmcget, et al: Wow... those are hard ones. I'll try, but unfortunately, a lot of this is "soon come":

1. What are the consumables cost per run?
We need another week to nail this down. There is also a nasty little catch-22: If we drop half a mil and move to flask scale, we can substantially lower the consumables cost, but we would also find ourselves sitting on what might be a year's worth of run kits. The biochemistry is moving fast, with multiple parties beavering away to take the current 26 paired end tag read length up to 35 or even 48 bases. If one of those succeeds a month or so after we funded a scale-up, it would be personally... embarrassing.

On the plus side, as detailed on the polonator.org site, we are already using a very low cost ligase, a license-free PCR polymerase, and license-free "Freedom Fluors". On some abstract level, we're "marching from Selma"!

2. How long does the DNA sample prep take?
I haven't had the, uhh, privilege of stepping through this personally, but I have heard that you should set aside about a week. If you didn't sleep, it would probably happen faster.

3. What is the minimum amount of starting material you can use?
The OpenWetWare Wiki suggests 50 ug, but there is an RCA step, and I'm not really sure how little would work. I'll check with Greg and get back on this. We can also make smaller flowcells.

4. What is the current sequencing accuracy of the machine?
We will have good data to answer this in about a week.

5. How many bases of mappable data can the machine generate per day and per run?
We will have good data to answer this in a few weeks (sorry). I think it is a big number.

Also any hints on what the spec of the A.008 will be?
We have some nice ideas, but since they could easily go down in flames, we would prefer to test them first. The "A" notion was an error on my part. The next revision will be H.008. But G is a fine thing, and all upgrades will be retrofittable, so come on in, the water's fine!

Kevin

[Chipper]

Hi Kevin,

I am working with ChIPseq data at the moment but from what I have read about the Polonator it seems even better suited for mapping enriched fragments, I think 2x13 bp shoud work well for this purpose. Do you have any news about approximate reagent costs per run, and what sort of output you get at the moment?

thanks,
Ola

[Kevin McCarthy]

Ola: We have approximate costs determined for the reagent kits, but we can't finalize these until our volumes are nailed down, and some licensing discussions are completed. The G version is undergoing performance testing now, which will allow us to pin down the reagent volumes. I would expect to be able to publish kit costs by the beginning of April. The initial costs should ramp down as we achieve scale.

Kevin McCarthy

[Chipper]

Any news from the performance testing?...

[Kevin McCarthy]

Ola: Things always go a little slower than one would wish... The initial G.007 Polonator delivered to the Church Lab in February underwent some serious tire-kicking. The Lab has very high standards. While most of the system was awesome, they found a few details that did need to be addressed. We beavered away on those, and delivered a fully updated G last Friday. At the same time, the Church Lab needed to complete their software, reflecting some of the improvements. They have now done so, and are just now beginning sequencing runs. These will yield data shortly, although this will be first pass, and won't be optimized for the platform for several weeks. We have also been making progress on licensing issues, although a firm out-of-the-gate reagent kit cost is now looking like the beginning of next month. There is also a real trade-off here between scale and cost. Costs will go down substantially once we begin to acheive scale, but we can't do this from day one. Bear with us; it will be worth it!

Regards,

Kevin

[horigen]

Ola: Things always go a little slower than one would wish... The initial G.007 Polonator delivered to the Church Lab in February underwent some serious tire-kicking. The Lab has very high standards. While most of the system was awesome, they found a few details that did need to be addressed. We beavered away on those, and delivered a fully updated G last Friday. At the same time, the Church Lab needed to complete their software, reflecting some of the improvements. They have now done so, and are just now beginning sequencing runs. These will yield data shortly, although this will be first pass, and won't be optimized for the platform for several weeks. We have also been making progress on licensing issues, although a firm out-of-the-gate reagent kit cost is now looking like the beginning of next month. There is also a real trade-off here between scale and cost. Costs will go down substantially once we begin to acheive scale, but we can't do this from day one. Bear with us; it will be worth it!

Regards,

Kevin

Means it's very possible for G.007 Polonator to get a $1000 genome?