Ion Torrent through the roof...

[james hadfield]

This article was referenced on GenomeWeb.

They comment that the Ion Torrent technology according to Jonathan Rothberg will 'improve 10-fold every six months'.

If they are getting 0.01Gb today then in eighteen months the PGM will be generating 100Gb and by the middle of 2013 it would throw out 10Tb of data per run.

My maths might be screwed up but that's a lot of data.

Anyone running one and posting yet?

[HESmith]

There's nothing wrong with your math, but maybe Rothberg's is suspect :-).

I, too, would be interested to hear results from someone with this system.

[NextGenSeq]

Similar to the 454 homopolymers are difficult for it. At a talk they showed it can get thru 8 or 9 As and they said it probably will never get through more than 15.

[ECO]

Moving to the new IonT forum!

[krobison]

Similar to the 454 homopolymers are difficult for it. At a talk they showed it can get thru 8 or 9 As and they said it probably will never get through more than 15.

For de novo sequencing and some resequencing, this is a major issue -- but then again, for a lot of applications it might be a tolerable flaw.

[wildung]

Anyone running one and posting yet?

Test fragments ran fine. Playing with our first libraries, now. Lots of people said it was all hype and asked why would anyone buy a 10Mb instrument. This is why. It always looked to me like it was going to end up with a) reads as long as the 454 once the phasing problem was solved and b) a greater number of reads by simply using modern semi-conductor feature densities instead of the low density features used for the 314 chip. But I'm still pleasantly surprised how fast both are moving forward. ABI did good.

Personally I'll take 1Gb of 300bp reads over 10Gb of 100bp reads any day, and I can't wait until short read assemblers are a thing of the past.

[james hadfield]

Joe Boland from NCI presented preliminary data at AGBT on Ion Torrent. They got their machine in January and presented at an international meeting in February. I'd say that say a lot about the system, and Joe's lab of course!

[SeqAA]

Similar to the 454 homopolymers are difficult for it. At a talk they showed it can get thru 8 or 9 As and they said it probably will never get through more than 15.

Biologically 6 homopolymers is really the metric. If they are already over 6 then they are ahead of 454 and have the bulk of the problems surpassed.

[wildung]

My $0.02

Got 15Mb out of the 10Mb chip 1st time out of the box and cut a ridiculous number of corners in the protocol. From breaking emPCR to starting the seqn run can be an easy one person day. Still would like automation for library prep, emulsion breaking and bead purification, preferably yesterday. The 100Mb chip is now out,, ahead of schedule, and a Gb by Fall,, there seems to be plenty of head room for growth and the technology appears at first glance to be robust and not nearly as fussy as some had feared.

[wraithnot]

Got 15Mb out of the 10Mb chip 1st time out of the box and cut a ridiculous number of corners in the protocol. From breaking emPCR to starting the seqn run can be an easy one person day. Still would like automation for library prep, emulsion breaking and bead purification, preferably yesterday. The 100Mb chip is now out,, ahead of schedule, and a Gb by Fall,, there seems to be plenty of head room for growth and the technology appears at first glance to be robust and not nearly as fussy as some had feared.

Great to hear that the machine works as advertised! What sort of read lengths did you get? Did the extra 5 Mb come from more sequence reads than expected, longer sequence reads than expected, or some combination of the two? Any idea about the error rates?

[wildung]

144k reads, 14.9Mb newbler says 77% Q40+ by consensus accuracy. >99% of my reads mapped to the refn. Not bad out of the block. Bet the next improvement in signal processing is will take the Q up to where it needs to go.

[wildung]

oops

consensus accuracy was 99.1% on a 5Mb genome.

[beardy_man]

Got 15Mb out of the 10Mb chip 1st time out of the box and cut a ridiculous number of corners in the protocol. From breaking emPCR to starting the seqn run can be an easy one person day. Still would like automation for library prep, emulsion breaking and bead purification, preferably yesterday. The 100Mb chip is now out,, ahead of schedule, and a Gb by Fall,, there seems to be plenty of head room for growth and the technology appears at first glance to be robust and not nearly as fussy as some had feared.

Sound like some promising data quality for a first real run I’d be interested to hear more about which corners you cut in the protocols. Will you be posting on the Ion Community site too?
Fully agree Re: emulsion purifications. This is a key throughput and hands-on consideration in comparison to the MiSeq. BTW, there are rumours that a single unit automation solution for the PGM is imminent (next few weeks), but has anyone tried joining up a REM e to the PGM emulsions in the mean time?

[wildung]

I would be shocked if ABI didn't jump on the automation problem before the transaction was complete. As for posting, I usually don't. I can't post too much because I really enjoy being this old and still having "junior member" status. But at this moment I feel I can contibute to anyone thinking about jumping in, or getting started w/emPCR for the first time.

Ideally, corners to cut in any proceedure are those that make it faster and better. For starters I washed the plate more, because I'm greedy, then stacked the aqueous in fewer tubes for less handling. I nearly eliminated the spins at both the non-polar and polar extractions. A DNA covered bead has so much charge that is isn't going to be anywhere near hexane. A hard 5' spin to separate the phases is unneccesary, we aren't trying to pellet the beads here. Anything more than a long pulse only puts junk you want to get rid of closer to the interface. There won't be anything you want at the interface, so it is OK to lose a ul or two of aqueous, to get the sample cleaner, quicker with fewer extractions.

The tiny beads are almost invisible, but seem to make a dense enough pellet with shorter spins. I was able to do more thorough washes by getting closer to the pellet when removing the super. It felt dangerous not spining long and getting all the super out, and it was, but it worked out fine. I melted the enriched beads off with 2x5' alkali soaks, instead of 1x15', etc. etc. etc.

The real fine line that needs to be walked, as with the 454, is the ratios going into emPCR. Only good quantification will help here. Aim low for fewer higher quality beads, then get greedy during their purification. Ideally you want to harvest just enough beads to proceed. All common sense stuff, wish it was worth more than $0.02, but it is all obvious.

[guz]

wildung, you should check out Ion's bulk breaking protocol. It is much quicker and far less tedious than spinning individual tubes.