[nickloman]

Where could they go with 454 that ion torrent is not already going?

Too late now, they should have done more back in 2009 (not just announce 1kb reads).

[GW_OK]

Exactly. The development of both SOLiD and Ion Torrent - both emPCR instruments - shows that they could have scaled 454 way further than they did, and also made steps to automate the fiddly library preparation (e.g. OneTouch). But they apparently chose not to, or couldn't for some reason. (No I don't count the REM e-System).

Yeah, I can't recall how many people I pestered at Roche about something called Raindance, that was going to revolutionize emPCR...

[GW_OK]

The irritating thing about this is that Roche could pour enough money into 454 development to make it competitive with Illumina. They apparently have not thought it worthwhile.

Where could they go with 454 that ion torrent is not already going?

And everybody poaches from everybody in the industry, particularly Illumina from ABI/Life. My Illumina sales rep tried to sell me on a SOLiD years ago, and some of the field service reps/managers worked on 3700's in my former lab.

[matholomew]

You are not making any sense whatsoever. Both IonTorrent and 454 do sequencing by synthesis (SBS)! The major problems with 454 sequencing are the optics required for detection and limited throughput. The IonTorrent chips were explicitly designed to address these limitation.

I agree the IonTorrent is just a 454 with a different sensor. It should have a slight advantage in the level of signal discrimination by using a pH meter versus a CCD - though not sure we've seen it. However while it's true Ion reads seem to be under Q valued, and Illuminas over, it looks like Ion reads ARE lower quality than Illuminas or 454 reads.

While pyrosequencing is a form of SBS it is different to Illuminas as it doesn't use reversible terminators so it has the same issues with homopolymers in 454 or IonTorrent. If you think it doesn't have an issue with homopolymers then I'd like to see it sequence some polyA tails.

Don't get hung up on semiconductors versus optics - it's all post light marketing. Flourogenic sequencing is more sensitive than either 454 or Ion method http://www.nature.com/nmeth/journal/...meth.1629.html. And both Helicos and PacBio show that you can sequencing single molecules with optics.

[nickloman]

The irritating thing about this is that Roche could pour enough money into 454 development to make it competitive with Illumina. They apparently have not thought it worthwhile.

Exactly. The development of both SOLiD and Ion Torrent - both emPCR instruments - shows that they could have scaled 454 way further than they did, and also made steps to automate the fiddly library preparation (e.g. OneTouch). But they apparently chose not to, or couldn't for some reason. (No I don't count the REM e-System).

[GW_OK]

You are not making any sense whatsoever. Both IonTorrent and 454 do sequencing by synthesis (SBS)! The major problems with 454 sequencing are the optics required for detection and limited throughput. The IonTorrent chips were explicitly designed to address these limitation.

Alright, if you want to get technical then yes, pretty much all of the NGS on the market use SBS. HOWEVER, the Illumina does a single iterative base for each synthesis step incorporating all 4 nucleotides at the same time, while the 454 and ion torrent flow each nucleotide across separately, watching for incorporation at each step.

And yes, the semiconductor methodology of the ion torrent does address the optical drawbacks of the 454 but it's still the same process overall. It's still going to have homopolymer problems. It's still going to rely on bead deposition on a plate or chip or whatever. It's still going to have the problem inherent with emPCR.

Finally, perhaps we could move the ion vs illumina discussion to another thread and not derail the discussion of the general dread caused by Roche trying to snatch up Illumina?

[pmiguel]

The irritating thing about this is that Roche could pour enough money into 454 development to make it competitive with Illumina. They apparently have not thought it worthwhile.

Illumina has a lot going for it, but where they seem unique to me is their level of tech support. Tech support from nearly every company I have ever dealt with tends towards to be close to a "nearly worthless" line. I presume this is "by design". Tech support is no doubt expensive and good tech support probably more so. Douglas Adam's was not far from the truth with his parody of this issue, with his depiction of the Complaints Department of the Sirius Cybernetics Corporation and their motto "Go Stick Your Head in a Pig". Seems like resisting the urge to cut tech support requires a sort of fortitude that larger companies just can't come by.

But I suppose there is also just a level of prestige that Illumina offers in the industry. They have plundered the workforces of their competitors, often luring away their best and brightest. But I suppose there is nothing like being bought by a diagnostics giant to turn a trend like that on its head. Could be worse. They could have been purchased by GE.

--
Phillip

[steinmann]

For research the $1000 genome is great. For clinical applications, you need to be looking at the $100 genome or maybe even the $10 genome if you have a public health service and are a trillion dollars in debt (like the UK is).

My experiences working in research is that the cost of sequencing is fine at the moment when using our Illumina machines. I'm not sure how many researchers would want to routinely do whole genomes as the analysis would be a bit challenging.

About 10-15% of the cost of any project we undertake comprise of sequencing. The remainder are split between staff and sample prep.
Bottom line: We need cheaper scientists and cheaper library prep kits!

Yes, a 100$ genome will be more awesome and have a larger clinical impact than a 1000$ genome.

Yes, the prices are pretty amazing right now compared to a few years ago, but not quite at the level where its something that you would do as casually and fast as a qPCR. Thats what I am looking forward to.

[steinmann]

The ion torrent is a 454 with a different sensor, and try as it might it's still going to run into the same problems and still going to be unable to compete with SBS.

You are not making any sense whatsoever. Both IonTorrent and 454 do sequencing by synthesis (SBS)! The major problems with 454 sequencing are the optics required for detection and limited throughput. The IonTorrent chips were explicitly designed to address these limitation.

[steinmann]

Hmm.. what is your opinion on the inefficiency of emPCR and the wells?

One should expect that between 30 and 50 % of the wells provide reads that pass the quality filter.

Quality values appear pretty good actually for existing chips. You have to consider that Ion Torrent underestimates Q values, while Illumina overestimate them. Someone from this forum did an independent analysis on this (PGM vs. MiSeq).

What will be really interesting to see is how making the wells smaller will affect signal to noise ratio and Q values. Thats really hard to predict.

[chadn737]

That's definitely wrong in that any of the available sequencers will sequence DNA from any organism (even arsenic-life ). But their marketing pitch is human exomes and human WGS, and any added value software might be tailored for those applications, and obviously exome capture kits will be for human samples.

Thanks for the correction. Like I said I am unfamiliar with their technology, our core does not have an Ion.That and the marketing for the Proton has been so biased towards human samples that I can't even really find mention of its use in anything else on their website. This mistakenly led me to believe that the chip was tailored for human sequencing only. I knew the PGM can do any sample, but it has a far lower output.

Discount my argument then.

[nickloman]

I'm not that familiar with Ion, but doesn't the Proton currently only work with human samples? Correct me if I am wrong. While the PGM does not have near the output of the Illumina.

That's definitely wrong in that any of the available sequencers will sequence DNA from any organism (even arsenic-life ). But their marketing pitch is human exomes and human WGS, and any added value software might be tailored for those applications, and obviously exome capture kits will be for human samples.

[chadn737]

I'm not that familiar with Ion, but doesn't the Proton currently only work with human samples? Correct me if I am wrong. While the PGM does not have near the output of the Illumina.

While the market for human genome sequencing is without a doubt the largest, keep in mind that there are a lot of other species that are worked on besides humans. I work with plants and for us Ion is not something we even consider when its time to do a next-gen experiment.

It seems to me that until Ion becomes a more viable option for other Eukaryotic species besides humans, there will still be a significant market for Illumina.

[GW_OK]

The ion torrent is a 454 with a different sensor, and try as it might it's still going to run into the same problems and still going to be unable to compete with SBS.

That being said, I really don't think having a $1k genome is going to be more than a technological milestone. No doctor is going to have somebodies genome sequenced except for some rare congenital syndrome. MAYBE cancer genomics, but still, it's not going to be that widespread.

The new big thing in sequencing is going to be clinical sequencing, I think we can all agree on that, but it's going to be custom gene panels or amplicon sequencing. A $1k test followed by a $10k analysis.

Getting us back on track, I shall cast in with those who oppose the merger on general principle, having grown up on a 454 before graduating to an Illumina system, and having seen the ponderous nature of the Roche R&D group.


Edit:
And has anyone seen ILMN's stock price since the announcement, up 44% to $53, Roche is going to have to ante up...

[NGSfan]

Think that this is the point were you are wrong. Clusters are not Illuminas killer tech. Its heir most serious limitation! Scanning billions of clusters is slow and takes expensive optics that are prone to failure.

Semi conductors are the killer tech here. They are much faster, scalable and so chap that you just trow your chip away after using it. Sure, we are not quite there yet. But you have to consider that Rothberg has kept all of his promises so far, shown us the machine and chips and will be shiping them to ealy access customers this spring. Ion Torrent will make the current illumina tech obsolete.

Hmm.. what is your opinion on the inefficiency of emPCR and the wells?

Note that when they claim something like 11 million wells, only 50% of them will contain all the reagents needed to carry out the reaction. They will have to scale enormously to catch up. Quality scores also don't look so good on the published runs.

Let's hope they can improve, otherwise I don't see Ion Torrent catching up.