Recently I posted a message asking for feedback from users of the Ion Torrent and how it compares to the Illumina MiSeq.I received many responses from others interested in the answer to this question, but unfortunately I did not receive any replies from anyone actually using this equipment. I did however receive some links to other sites with pertinent information and a very informative comparison of these two platforms.These are listed below.I would still appreciate any additional information on this topic. Thank you, Carla Hurt, Ph.D. hurtc@bio.miami.edu Dept. of Biology Cox science center, rm 36 1301 Memorial Dr., University of Miami Coral Gables, Fl 33146 1.) Here are some blog posts that you might find helpful: http://www.massgenomics.org/2011/02/small-ball-sequencing-on-the-benchtop.html http://www.massgenomics.org/2011/04/iontorrent-benchtop-sequencing-streamlined.html http://www.massgenomics.org/2011/06/first-look-data-from-iontorrents-316-chip.html And then here is a recent anouncement from Life Technologies about a new small sequencer: http://goo.gl/9X8lc 2.) Here is a very helpful blog with information about many NGS platforms: http://omicsomics.blogspot.com/ I also contacted Dr. Keith Robison - the author of this blog - and asked him specifically to compare the Ion Torrent and the Illumina MiSeq.His response is pasted below. Just as a reminder -- I haven't had hands on either machine (I have yet to see a MiSeq up close) but I have had data generated for me on both systems & have tried to seriously envision what each system would be like to have in a lab at my employer in terms of cost & labor. Some key points to consider: Both machines have a sequence output well suited for amplicon sequencing and for small genome sequencing (probably up through a few tens of megabases). Neither is well suited for really large genomes, metagenomes, for transcriptomes or for counting applications (e.g. ChIP-Seq) -- the numbers of reads and amount of data are really too small and the cost per basepair is much higher than a system like HiSeq. From a user's perspective, the main two reasons to go with these systems rather than HiSeq/GAIIx is the low cost per experiment and rapid turnaround time. MiSeq has a greater range of software on the academic/freeware side; Ion doesn't really seem to have caught the bioinformatic developer community's imaginaition. Some software developed for 454 should work with Ion Torrent data, due to each having similar error profiles (homopolymer runs); this error profile also frustrates some software designed for Illumina. On the other hand, a lot of tool are platform-agnostic. On the commercial side, there does seem to be healthy support for all platforms. Contrasting the two systems further are the read profiles. MiSeq offers a number of read lengths up to 150x2 paired end (which is longer than HiSeq and perhaps one reason why MiSeq may have an edge over HiSeq for genome assembly). Ion now has ~250bp reads (you get a distribution) and a paired end protocol, though not a lot of groups are using it & I haven't seen 250x2 paired end data yet (which would be very cool to have!). Both are probably going to have further read length improvements. I'm guessing the next Ion mark will be in the 300-350 range and probably be released sometime this spring. Several groups have demonstrated greater than 150 bp runs on the MiSeq, and Illumina will be releasing longer kits this summer (I think 250x2 is the first one). Overall, I think the true end-to-end run times are getting very similar, but Ion may still be a bit better -- if you are willing to put in the extra labor. Also, in planning throughput the timings can be troublesome. For example, right now the 150x2 mode takes 27 hours, which means if you were running just that & didn't have shifts 24x7, it is hard to keep the machine cranking all week (Illumina is finding clever ways to shorten cycle time; I think their reps told me the 250x2 will take /less/ time than that, though I could be remembering that wrong.). it does appear that the 318 chip is the end of the PGM line, though LIFE has not made that explicit -- Proton will be the instrument for doing big sequencing. MiSeq will get an upgrade this summer to allow reading both surfaces of the flowcell; this will roughly double the amount of data. Both will probably top out at a size suitable for targeted sequencing in model systems or with very focused exome arrays (but not whole exome). For owning one, MiSeq is a much less labor-intensive system once a library is prepared. You basically just shoot it into the machine and it goes. Paired end protocol is handled on machine; no further intervention is required. In contrast, even with the OneTouch the protocol for going from a library to template for the machine requires several manual steps. If you are running paired end, it requires further manual intervention. For library preparation, MiSeq has Nextera. This obviates the need for a Covaris or similar shearing instrument and can work with small input quantities. However, it is very sensitive to the amount of input DNA & we've had two very disappointing Nextera libraries that were mostly ultrashort inserts -- our DNA concentration was overestimated by a sizable amount. We just got a Pippin Prep sizing instrument & will make our own libraries in the next few weeks -- I think the size selection will add a lot. Ion does have an enzymatic protocol, but all the data I've gotten has been physically sheared. For amplicons, the Illumina platform has an Achilles heel which can be dealt with, but can be a nuisance. On Illumina instruments, a low complexity of sequence early on in a read fouls up the cluster finding. So you must pick barcode combinations with care & try other tricks to prevent this. Ion doesn't have this problem. On cost, Ion is a bit cheaper -- but the real cost is a PGM+ the server + the OneTouch, which I think sums to around $90K; MiSeq is more like $125K. One PGM server can handle 2 machines, so if you are getting multiples the math favors MiSeq a bit more. However, you really do need about 1:1 OneTouch:PGM ratio, and actually I could envision wanting more OneTouch instruments than sequencers might be more efficient for labor (but haven't fully worked that out; we couldn't imagine having multiples in our current space). Running costs are pretty similar, I think. Since I'm going through various vendors, that gets mixed with other costs & so I can't parse them out. -- Carla Hurt, Ph.D. Dept. of Biology Cox science center, rm 36 1301 Memorial Dr., University of Miami Coral Gables, Fl 33146 PH# (305) 284-6566 Core Lab server access: http://zorki.bio.miami.edu/~corelab/ Carla Hurt