[GW_OK]

How the heck does one do SMS with their current hardware geometry just by switching chemistry? Something does not add up.

I rather think it's a code name for something completely different.

[simonandrews]

Any hints as to whether there might be an upgrade to the HiSeq 1000 to provide the sorts of improvements seen in the 2500?

[ymc]

The throughput of HiSeq 2500 is double that of HiSeq 2000 because it can do 600Gb in 5.5 days as oppose to 11 days??

The only pluses are turbo mode and no need for cBot?

[BBoy]

Ok, so this isn't a LOT of new info, but here's something from the latest William Blair report (https://www.rdocs.com/GetRDocNoLogin...40&zID=34808):

"Although not too much detail was provided, we get the idea that
one chemistry features a fast cycle time (10 seconds per cycle) and long read length, while the other chemistry, code
named “Chemistry B,” features single molecular sequencing, which, according to Mr. Flatley, has high accuracy rate
and low cost, and could become a backbone of a low‐cost machine."

How the heck does one do SMS with their current hardware geometry just by switching chemistry? Something does not add up.

[scbaker]

Chemistry B

Good points.
Whatever Chemistry B is, semiconductor or no, I hope we see more of it soon.

Ok, so this isn't a LOT of new info, but here's something from the latest William Blair report (https://www.rdocs.com/GetRDocNoLogin...40&zID=34808):

"Although not too much detail was provided, we get the idea that
one chemistry features a fast cycle time (10 seconds per cycle) and long read length, while the other chemistry, code
named “Chemistry B,” features single molecular sequencing, which, according to Mr. Flatley, has high accuracy rate
and low cost, and could become a backbone of a low‐cost machine."

[winegard]

Just heard from our rep. Another interesting tidbit is no more need for a c-bot. Cluster generation is done on the system a la the miSeq with the 2500.

If you were thinking of buying a 2000 plus a c-bot, a 2500 looks like a pretty good deal with a $50k price differential.

[GW_OK]

Good points.
Whatever Chemistry B is, semiconductor or no, I hope we see more of it soon.

[GenoMax]

There is a big gap between doing "cool" science and ultimately fabricating a viable instrument that can survive rigors of use in less than perfect conditions.

[BBoy]

Illumina's optioned whatever Oxford Nanopore can come up with. Oxford Nanopore has licensed stuff that Golovchenko and Lieber have come up with. Lieber has recently published an interesting paper on nanopore sequencing in Nature Nanotechnology.

I think things might be further along than people imagine.

As a semiconductor geek I am willing to bet a good dinner that this is NOT what Oxford is looking to commercialize . The effect is interesting science, but combining silicon nanowire FETs with synthethic nanopores is well beyond the capability of even leading edge companies today with a commercially viable process. Heck, each one of these two critical ingredients by itself is out of reach for anything resembling mass production at the dimensions required for single-base resolution. I am also familiar with Lieber's body of work over the years, sometimes I kid that he has yet to see a nanowire or a nanotube he does not like. He typically publishes a few papers that utilize the high surface/volume ratio, then moves on to the next nanowire du jour. It is good science, but is invariably too far out from a technological perspective.

As the same semiconductor geek I hope I am wrong, this would be exciting stuff if I am wrong. I read on some blog that there is an expectation that Oxford will soon come out with a statement that they have been able to sequence a genome. We shall know soon enough, exciting times for us gearheads.

[GW_OK]

Yeah, this is what I am interested in too, don't have a pro subscription to Genomeweb and they seem to be the only ones that have commented on the SMS thing.
...

Illumina's SMS strategy is probably closely linked to whatever progress Oxford Nanopore is making, Illumina is an investor there.

Illumina's optioned whatever Oxford Nanopore can come up with. Oxford Nanopore has licensed stuff that Golovchenko and Lieber have come up with. Lieber has recently published an interesting paper on nanopore sequencing in Nature Nanotechnology.

I think things might be further along than people imagine.

[ymc]

So what is the latest per genome reagent cost for this new HiSeq 2500 machine?? Finally sub $1000??

[athos10025]

I don't think that there will be much more in reagent costs on a per-run basis. On the other hand perhaps the machines would be run more frequently. Most of our runs are multiple-sample projects. We often have 6 PIs spread across the lanes all with multiple samples. This can be hard because all of the sample arrivals and sample prep need to be in, more-or-less, in lock step. If some samples don't arrive or don't pass QC then the entire run get puts on hold. Or we eat the costs and do a partial run. On the other hand if we could just do a 120 GB at the same cost and with shorter turn around then we would put fewer samples in the run thus having fewer problems.

The SOLID 5500 allows for partial running -- the flowcell can be partially used and then wait around for the next batch of work to be prepared. That flexibility with a high-end machine is very nice.

Agreed. Illumina talked about this as a feature for clinical applications, but I think that this flexibility will be very much appreciated in academic cores.

[BBoy]

I've not heard anyone else talk about "chemistry B" or single molecule sequencing from ILMN. Can you share more?

Yeah, this is what I am interested in too, don't have a pro subscription to Genomeweb and they seem to be the only ones that have commented on the SMS thing.

GW_OK, I seriously doubt this is related to the article you linked to. Much of the work at Harvard (Gene Golovchenko?) historically has been around sensing with tunneling currents. The translocation speed of DNA through the pore needs to be slowed down substantially to have a hope of single-base detection. This can probably achieved with appropriate biasing. Even then the vertical extent of solid state nanopores that can be fabricated in mass quantities with practical means today is such that stray fields above/below the nanopore average over as much as 10 bases, which makes it hopeless to deconvolute the original signal. The envisioned workaround is to use nanoelectrodes with nearly atomically sharp tips. However, this makes them susceptible to the spatial orientation of the DNA as it translocates. Furthermore the currents that the electrodes need to sustain are such that they are destroyed in relatively short times. None of this is to say that at some point in the future this technology will not become viable, but at the present time it is far away from commercialization, I belelive.

Illumina's SMS strategy is probably closely linked to whatever progress Oxford Nanopore is making, Illumina is an investor there.

[westerman]

+1 krobison.

I will be very interested to see the per base pricing on the "Turbo" mode versus the "standard" longer run times. Clearly ILMN needs to get people stuffing more reagents into these big boxes...10-25 runs per year (as probably happens with most HiSeqs now) can't support a high growth business.

I don't think that there will be much more in reagent costs on a per-run basis. On the other hand perhaps the machines would be run more frequently. Most of our runs are multiple-sample projects. We often have 6 PIs spread across the lanes all with multiple samples. This can be hard because all of the sample arrivals and sample prep need to be in, more-or-less, in lock step. If some samples don't arrive or don't pass QC then the entire run get puts on hold. Or we eat the costs and do a partial run. On the other hand if we could just do a 120 GB at the same cost and with shorter turn around then we would put fewer samples in the run thus having fewer problems.

The SOLID 5500 allows for partial running -- the flowcell can be partially used and then wait around for the next batch of work to be prepared. That flexibility with a high-end machine is very nice.

[scbaker]

Chemistry B?

For those of you who haven't been monitoring the latest thread over in the Ion Torrent forum. From Twitter:
Illumina announces #HiSeq 2500: able to seq entire genome or 20 exomes in 24h. #JPM12 #GenomeInADay

Illumina introduces #MiSeq enhanced performance w/ 3-fold throughput increase (7 Gb), faster cycles & 250 bp PE reads. #JPM12


How exciting!


Edit:
@illumina is working on new chemistry. "chemistry B" which is their stab at single molecule detection. #JPM12

I wonder if this has something to do with this:
http://news.harvard.edu/gazette/stor...ilding-blocks/

I've not heard anyone else talk about "chemistry B" or single molecule sequencing from ILMN. Can you share more?