Announcement

Collapse
No announcement yet.

Ion Torrent PGM vs Illumina MiSeq

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • Originally posted by vehuardo View Post
    Hi all!


    My institute will purchase a benchtop sequencer for bacterial genome sequencing. It will be used both for routine genome seq and (importantly) for infetious disease outbreak analyses. Hence, although only one of many important parameters, time from DNA isolation to results is important.

    I am positively inclined towards the MiSeq, as I have only good experience with Illumina data (HiSeq and MiSeq). However, at a presentation at the institute, they weren´t really that impressive. Ion Torrent, on the other hand sent an expert on bacterial genome sequencing, who, it seems managed to sway a lot of people.

    I find it difficult to make a sound comparison. The published literature largely favors MiSeq for quality of sequences and throughput. However, Ion Torrent claims that the Ion 318 v2 Chip yields improved quality. And throughput isn´t necessarily that important. They also claim that library prepping is fast (<2 hrs) and automated.

    Also, what convinced people at my institute was data handling: the Ion Torrent seems to come with built-in software that can output variant calls and that can be tailor maid for various outputs (e.g. resistance gene mutations etc). This sounds nice, but I have a feeling that this way of doing things is not very flexible…

    Some specific questions:

    -Has anyone experience with long read MiSeq (2x250 or so) AND Ion torrent (418 bp reads)? Any thoughts on whats best?

    -Is mate-pair sequencing on Ion Torrent something that must be done in a separate run following «normal» single end sequencing?

    -Even though Ion Torrent is a lot more bulky and has more steps, It seems like it is a lot faster than MiSeq long reads. Is that correct?

    -Is it problematic doing comparative genomic analyses with data from Illumina and Ion torrent together?

    -M. tuberculosis (65% GC content) is among the bacteria that will be sequenced routinely. Ion Torrent seems to suck at high GC but I have only found one paper on this. How much is this of a problem? Ion Torrent showcased the use of Ion Torrent for quick resistance determination in M. tuberculosis, but this scheme relied on amplicon sequencing, which in my view is a bit like walking backwards into the future……

    Please excuse this chaotic post, but there are so many questions...

    All feedback appreciated,

    Vegard
    Well, Illumina has PCR free kit for metagenomics. If you are serious about metagenomics, Illumina is the only way for now.

    Comment


    • We bought the MiSeq when the PGM came out and was still comparable with the MiSeq. I don't know about Ion, but it seems the MiSeq simply works... consistently and as advertised. I've read in other reviews the Ion prep is very tricky but I'm not sure if that still applies with their new kits and machines. I would also take into consideration that there is a chance Ion won't survive given that they have fallen behind in all their promises. There are few niche applications where the PGM is better than the MiSeq, I'm not sure if these are enough to sustain their market, make sure that you don't buy a machine that will be discontinued in a few years.

      Comment


      • Originally posted by aleferna View Post
        We bought the MiSeq when the PGM came out and was still comparable with the MiSeq. I don't know about Ion, but it seems the MiSeq simply works... consistently and as advertised. I've read in other reviews the Ion prep is very tricky but I'm not sure if that still applies with their new kits and machines. I would also take into consideration that there is a chance Ion won't survive given that they have fallen behind in all their promises. There are few niche applications where the PGM is better than the MiSeq, I'm not sure if these are enough to sustain their market, make sure that you don't buy a machine that will be discontinued in a few years.
        Thanks for the info!

        Comment


        • Originally posted by vehuardo View Post
          Hi all!


          My institute will purchase a benchtop sequencer for bacterial genome sequencing. It will be used both for routine genome seq and (importantly) for infetious disease outbreak analyses. Hence, although only one of many important parameters, time from DNA isolation to results is important.

          I am positively inclined towards the MiSeq, as I have only good experience with Illumina data (HiSeq and MiSeq). However, at a presentation at the institute, they weren´t really that impressive. Ion Torrent, on the other hand sent an expert on bacterial genome sequencing, who, it seems managed to sway a lot of people.

          I find it difficult to make a sound comparison. The published literature largely favors MiSeq for quality of sequences and throughput. However, Ion Torrent claims that the Ion 318 v2 Chip yields improved quality. And throughput isn´t necessarily that important. They also claim that library prepping is fast (<2 hrs) and automated.

          Also, what convinced people at my institute was data handling: the Ion Torrent seems to come with built-in software that can output variant calls and that can be tailor maid for various outputs (e.g. resistance gene mutations etc). This sounds nice, but I have a feeling that this way of doing things is not very flexible…

          Some specific questions:

          -Has anyone experience with long read MiSeq (2x250 or so) AND Ion torrent (418 bp reads)? Any thoughts on whats best?

          -Is mate-pair sequencing on Ion Torrent something that must be done in a separate run following «normal» single end sequencing?

          -Even though Ion Torrent is a lot more bulky and has more steps, It seems like it is a lot faster than MiSeq long reads. Is that correct?

          -Is it problematic doing comparative genomic analyses with data from Illumina and Ion torrent together?

          -M. tuberculosis (65% GC content) is among the bacteria that will be sequenced routinely. Ion Torrent seems to suck at high GC but I have only found one paper on this. How much is this of a problem? Ion Torrent showcased the use of Ion Torrent for quick resistance determination in M. tuberculosis, but this scheme relied on amplicon sequencing, which in my view is a bit like walking backwards into the future……

          Please excuse this chaotic post, but there are so many questions...

          All feedback appreciated,

          Vegard
          Hi Vegard,

          I've been using both platforms for a while now and I'd say they are good at different things. Compared to the MiSeq, the PGM has lower max throughput, period. However the throughput of the PGM can be scaled depending on the chip you use. Currently there are three different chip sizes (314, 316 and 318). The 314 chip is the cheapest but also gives you the least amount of reads. We found this very useful as we would receive a small number of very high priority clinical samples that would not, by themselves fill a MiSeq run, thus wasting reagent.

          When it comes to sample-prep, there's the IonChef that automates a lot of the pre-sequencing workflow, we've had it for a while now and it works really well. The only downside is that it requires 2 chips per run, so either you'll have to run 2 chips concurrently (not necessarily with the same libraries mind you) or have 2 PGM's. The library prep using the Chef is also straighforward, minimizing user error. There is also a new chemistry called Hi-Q that decreases error rates and increases overall sequence quality.

          The IonTorrent server that takes care of the data can indeed be made to do variant calls and similar, generally we preferred to download the data instead and use CLC for mapping and variant-calling so i don't have that much experience with the inbuilt one. The server does support plugins though, so you can easily write your own pipeline in perl/python and make it run on your data directly on the server.

          -I'm using both, mainly you see a difference in error profiles, the PGM is more inclined at having troubles in homopolymers while the MiSeq is more evenly spaced out.

          -Mate pair on the PGM is fairly new, but it does not require an extra step, the sample prep is more complicated though.

          -Running time varies a bit with what kind of chip you use, but usually its something like 6-10 hours after loading it on the instrument. The chef requires an additional 10-12 hours if i remember correctly, so we usually loaded the chips into the chef before we went home and then ran the PGM the morning after.

          -When combining datasets, the different error profiles could mess with your analysis, there are a few read-error correction tools available (Hammer from the SPAdes pipeline for e.g.) that could help.

          -We routinely used the PGM for bacterial sequencing (Listeria and other food-borne diseases) but also for parasites and similar. The PGM performed admirably at both tasks, from 30% GC to 65% as far as we could tell.

          Bottom line:
          If you have a large volume of samples that arrive continuously (so you can maximize the throughput of the MiSeq) and turnaround time is not a critical aspect, the MiSeq is probably the better choice. If you have a varying amount of samples that need to be turned around quickly, you are better of with a PGM.

          //tracer

          Comment


          • Thanks Tracer,

            This answers all my questions. Brilliant!

            Regarding this:

            "Bottom line:
            If you have a large volume of samples that arrive continuously (so you can maximize the throughput of the MiSeq) and turnaround time is not a critical aspect, the MiSeq is probably the better choice. If you have a varying amount of samples that need to be turned around quickly, you are better of with a PGM"

            .. I guess we have a little bit of both, so we have to choose whats most important. Or buy both...

            Cheers,

            Vegard

            Comment


            • Any update o new thoughts about this thread ?

              Comment


              • Updates on Illumina vs IonTorrent

                a must read: http://omicsomics.blogspot.se/2015/0...uzz-again.html


                I have to agree with Keith here, I would love to hear some company giving Illumina a run for their money. The problem I see with torrent is that they made too many promises and failed to deliver, badly. Now they are restricted to a few niche markets, but is this enough to sustain development at the level needed to compete with Illumina?

                Comment


                • Originally posted by AntonioRFranco View Post
                  Any update o new thoughts about this thread ?
                  Wait for the minION.

                  Comment


                  • Hi tracer,

                    You are using PGM for bacterial identification? which gene target are you using? Are you using non-16S gene as well? or you using 16S with different V regions? what is your primer set to identify bacteria? What is your turnaround time for bacterial ID?
                    Thanks

                    d00b

                    Comment


                    • Originally posted by Chipper View Post
                      Wait for the minION.
                      It's out already. They are projecting over 40 GB of data for 1K

                      Comment


                      • Originally posted by NextGenSeq View Post
                        It's out already. They are projecting over 40 GB of data for 1K
                        Well, for values of "out" that include not being able to just order one.

                        Where do you get the 40 GB of data figure?

                        --
                        Phillip

                        Comment


                        • I agree that it's not really 'out' yet - so far the 40Gb claim comes from ONT but not any end users. I think this is from the MkI version that maybe isn't quite yet available to MAP participants. Does anyone have the latest 'real world' results from MAP participants? Last I saw was still sub-Gb per run.

                          Hopefully ONT will reveal more at their "London Calling" event (and hopefully some attendees with tweet about it or post the info here...)
                          AllSeq - The Sequencing Marketplace
                          [email protected]
                          www.AllSeq.com

                          Comment


                          • Originally posted by pmiguel View Post
                            Well, for values of "out" that include not being able to just order one.

                            Where do you get the 40 GB of data figure?

                            --
                            Phillip
                            Clive Brown's blog

                            Comment


                            • I've done a price/technology comparison of Illumina MiSeq and IonTorrent PGM here. IMO it's got to be illumina...

                              Comment


                              • all right, our lab is doing very similar jobs with vehuardo's lab, plus a lot of viral identification. And the major question confused me alot is the choice between S5 and miseq.

                                our fund is only supported for one miseq, so it's hard to buy S5+ionchef. any information will be welcome, thanks!

                                ps: like vehuardo said, the Life tech. did a good presentation that are impressive most of my collegues.

                                Comment

                                Working...
                                X