Ion Torrent $1000 Genome!? Benchtop Ion Proton Sequencer

[aeonsim]

Does any one have any more information on this? Or had one in testing under an NDA and can now talk about it?
Looking at the press release it seems they're claiming they'll have the whole Genome chip available near the end of the year.
Looking at the images it looks like they're using a newer much larger chip.

Does any one know what coverage they're talking about when they say a $1000 Genome, 1x, 10x or 30x?

http://www.lifetechnologies.com/glob...-io-proto.html

http://www.invitrogen.com/site/us/en...ng/proton.html

[ECO]

Way to catch that immediately!

The only details that can possible be extrapolated from are in this quote:

The Ion Proton™ Sequencer is ideal for sequencing both exomes — regions in the DNA that code for protein — and human genomes. The Ion Proton™ I Chip, ideal for sequencing exomes, will be available mid-2012. The Ion Proton™ II Chip, ideal for sequencing whole human genomes, will be available about six months later. In addition, the Ion Proton™ OneTouch™ system automates template prep and a stand-alone Ion Proton™ Torrent Server performs the primary and secondary data analysis.

"Just six months after our first semiconductor sequencing chip was released, people used it to solve the German E. coli outbreak, sequencing the toxic strain in just a couple of hours," said Dr. Jonathan M. Rothberg, the Founder and CEO of the Ion Torrent division. "Now, six months later we're developing a chip that's 1,000 times more powerful than that to sequence an entire human genome in about the same amount of time. That's the power that semiconductors bring to sequencing."

Extrapolating from above:

chip	sensors	tput	availability
First chip (314)	1.3 Millon	10MB	Mid 2011
1000x	1.3 Billion	10GB	early 2013

[ECO]

Image of the thing added to your post...and moved to front page.

[scbaker]

Originally Posted by ECO

Way to catch that immediately!

The only details that can possible be extrapolated from are in this quote:



Extrapolating from above:

# of sensors (reads) is only one dimension that can be increased. The other major one is read length. I was a little surprised to read in a previous press release that they won't officially launch 400b reads until the second half of 2012 (unclear if that's to be on both the PGM and Proton).

[ECO]

Another pic of chip...



and of their R&D test rack. I wish that guy would get out of the way...

[aeonsim]

There's a bit of additional data in this article: http://www.genengnews.com/gen-news-h...-day/81246184/

Which if they've quoted them correctly the Proton II chip has 660M sites, so in that case it may be that half the increase is coming from the additional sites and the rest from read length?

[ECO]

Originally Posted by scbaker

# of sensors (reads) is only one dimension that can be increased. The other major one is read length. I was a little surprised to read in a previous press release that they won't officially launch 400b reads until the second half of 2012 (unclear if that's to be on both the PGM and Proton).

I'm sleep deprived, thanks.

Yes, let's give them 500bp read lengths instead of 100bp on the 314. So ~260M active sensors? Either way I read "1000x more powerful than the 314" and start to do estimates from there, maybe after more sleep.

[ECO]

165M for Proton I, which they call Exome scale
660M for Proton II, which they call genome scale

Lots of broad assumptions in here...but seems roughly doable without much movement in readlength.

[aeonsim]

Originally Posted by ECO

Way to catch that immediately!

The only details that can possible be extrapolated from are in this quote:

Extrapolating from above:

Hopefully we'll be looking at more than 10GB of sequence. While ~3x Coverage of the genome is technically a $1000 genome I'd think 10x-30x would be a level of coverage far closer to what most people would consider a $1000 Genome.

If we say 660M sites as the third article claims, and a read length of 200bp, then a quick of the back of my hand calculation gives something like 30x coverage of a human genome at 70% of the sites being active.

[nickloman]

I've summarised the news here http://pathogenomics.bham.ac.uk/blog...t-the-machine/ and added a few thoughts of my own.

[454newbie]

What % sensors are active on the current generation of chips, and what % are occupied with beads. I thought they were still dependent on poisson distribution, so loading at < 30% of total sites.

Also- what are the implications for pixel (and bead size) with the Proton chips?
The chip looks to be about 15mm by 2mm (unless those are giant or midget hands), or 15000um x 20000um, giving a total area of 3x 10^8 um2. 170x 10^6 active sites implies only ~2um2 per pixel. This, in turn suggests feature sizes of no more than .5um2 (allowing for spacing between features).

It seems possible that this is now no longer going to be a bead-based platform...but with beads replaced by colonies, clusters or some such thing.

[nickloman]

Some users (including ourselves) are managing to get >75% live wells on the 316 chip or even higher.

I think this will still be a bead-based system (but smaller beads, wells and spacing), but it's an interesting question!

[nickloman]

BTW I would think the chip is more like 20mm^2 (old chips 9mm^2) and they will be working on near-complete loading.

[TonyBrooks]

No wonder Life are selling the original PGM's off at a dirt cheap price.

[nickloman]

Originally Posted by TonyBrooks

No wonder Life are selling the original PGM's off at a dirt cheap price.

I am interested to know how Ion Torrent PGM customers feel about this news...

[GW_OK]

Pixels? I didn't think the ion torrent needed pixels?

Also, it seems like right now they've got a machine equivalent to a single hiseq lane costing slightly less if only reagents are being taken into account. Running quicker, of course. But is it better, faster, cheaper than a Miseq?

I'm also slightly confused by the genomeweb article, where the Proton won't leave people with a mountain of data like the hiseq? And yet it still produces what should be tons of data?

Furthermore are read lengths >4-500bp feasible, really?

[turnersd]

Yeah I wouldn't be very happy unboxing a new instrument with a planned obsolescence date of mid-2012.

[nickloman]

Originally Posted by GW_OK

Also, it seems like right now they've got a machine equivalent to a single hiseq lane costing slightly less if only reagents are being taken into account. Running quicker, of course. But is it better, faster, cheaper than a Miseq?

Right now they haven't got anything as this is just an announcement... first machines to early access customers mid-2012.

MiSeq spec as of now is 1.5-2Gb per 27 hours (2x150bp).

We've extrapolated that Ion is looking at maybe >20 Gb per run (3 hours?) for Proton 1 chip in Q3 2012, and >100Gb per run for Proton 2 Q1 2013. So it promises far greater throughput and faster than the MiSeq.

Currently MiSeq is cheaper than the Proton but more than the PGM.

Strongly expect to see Illumina announcements reacting to this!

Furthermore are read lengths >4-500bp feasible, really?

We don't really know!

[TonyBrooks]

Originally Posted by nickloman

Strongly expect to see Illumina announcements reacting to this!

Obviously not a good time for Illumina to inform us that our sequencing prices are going up by 10% then.

[nickloman]

Originally Posted by TonyBrooks

Obviously not a good time for Illumina to inform us that our sequencing prices are going up by 10% then.

Hmm perhaps not, but Illumina has >90% of the sequencing market right now.