Thursday, February 8, 2018

How fast is an Orbitrap Fusion 1 with the "30Hz" upgrade in software 3.0SP1?


I swear, I'm not writing this post to annoy people. This is important information for those of us trying to design the best possible experiments and it is info that can sometimes be hard to get. As I continue to shorten my runs to get the highest possible throughput in samples and replicates per day out of my instruments, the scan speed of each instrument is of paramount importance to my quantification accuracy (I want 12 measurements across each peak -- minimum -- that's why I spend so much time working on "how fast can I get it")

In the newest Fusion software there are some fantastic new features. I 100% recommend doing the upgrade. I...um...didn't have the most efficient upgrade path doing it on my own -- my friendly neighborhood FSE suggests operator error and I don't care, it needed a PM anyway, --what I do care about is the Fusion has been absolutely killing it since the upgrade and cleaning!

One new feature is the ability to run the Orbitrap at 7,500 resolution. The release notes describe it as "30Hz under specific conditions" and this is absolutely the case.

To see how fast the instrument can run, I loaded 200ng of a complex human cell digest. I set the fill times at increasingly lower intervals. From 30ms all the way down to 10ms total fill time for the MS/MS.

With a 10ms max fill time the space between MS/MS scans is right around 30Hz.


Want to know the surprising part? This run isn't even complete garbage! 10ms max fill with only 200ng on column? You do that for experimental purposes just to see how fast the instrument is -- in theory -- you don't expect IDs. And...I still pulled almost 1,000 unique protein groups on this run! Raise this to 22ms and the number of ID's I pull are a good bit better, though!

Okay -- so the Orbitrap can do 30Hz "under specific conditions" as advertised, honestly it is a little better than this! Right on. However, one of the things we've seen really emphasized in the literature -- for example,  in this study and this study and this study is the concept of sequencing speed as a measurement of what is fully achievable in a standard LC-MS/MS data dependent experiment.

The Fusion 1 with the new 30Hz upgrade can achieve a ddMS2-MS2 sequencing speed of 30Hz if you run the MS1 at 7,500 resolution as well. If, however, you wish to use a higher resolution you're going to take a hit in the overall sequencing speed.

I generally do my LFQ runs with at least 120,000 resolution at MS1. Might be overkill, based on 3 studies I've really liked this year, but that's how I'm going. This is around 4Hz. If we run the instrument with a TopN or use a TopSpeed method with 1 second then that 4Hz scan is going to have a massive impact on the overall sequencing speed.

This is where the math gets tricky (I'm working on it for a possible update of the Quadrupole Orbitrap Cycle time calculator...we'll see...). If you are using a Top speed method that doesn't require an MS1 scan every second, the hit in Hz isn't nearly so bad. For example, the proteomics methods seem to have a default of TopSpeed=3 seconds.

Let's check this math. If you run a 120,000 MS1 scan on the Fusion 1 (assume no overhead, so 4 Hz) and assume a 33ms time to complete the 7,500 resolution MS/MS scan -- if you do a MS1 scan every 1 second, it looks like this:


You are getting around 22-23 scans/second in the Orbitrap. This is comparable to the Q Exactive HF; albeit I am getting 2x the resolution at the MS1 (critical to me) and I'm getting half the resolution at the MS/MS (which I am totally fine with!)

However, what if you do only have an MS1 scan every 3 seconds?


You are going much faster than the Q Exactive HF. We're looking at >27 MS/MS scans per second. Niiice! This is what I'm looking for.

Now...it's still a far cry from...


...but you probably got a tribrid system because you wanted an ion trap. It is worth noting that this recent study states that the ion trap on the Fusion II system exceeds 60Hz when using the ion trap in conjunction with the Orbitrap.


...so if speed is your #1 concern you can always use that second mass analyzer!

7 comments:

  1. I'm not an expert but I think you've forgotten about the transfer time of ions from HCD to OT about 10 ms. So your reasoning is also applicable to resolution 15000 :(Tune 2.0), MS2 OT 15000 resolution with injection time 22 ms.

    ReplyDelete
  2. Ben did you intend to write HeLa in lieu of complex human cell digest? If not how complex vs HeLa is your sample? Also, would you share your chromatography(gradient time, column length, injection volume)?

    ReplyDelete
  3. Andrea, thanks for the comment. I am considering the HCD fragmentation time as part of the "overhead". Is HCD 10ms? I'll look into it. That seems really slow, if true!

    ReplyDelete
  4. Andrew,
    Yeah -- I'm using the HeLa -- for now. I dislike using anything from that cell line, but I wanted to match historic data and it's the best one to start with. We won't be using it in the future, or for the publication we're preparing.

    ReplyDelete
  5. Hi Ben.

    SP1 is that Orbitrap Fusion version 3.0.2022.16? It has the 7500 and 50K resolution option for TMT. Where do get your HeLa digests or cell lysates? Do you buy it digested from Thermo? Or prepare your own digests?


    _Darryl (dhardie@proteincentre.com)

    ReplyDelete
  6. Darryl,
    Yup! That's the one, I think. For this I bought the HeLa digests because I wanted to match historic data from other instruments. It won't be our QC spec material, but it was helpful in months 1-2 at my new job to get a baseline for where the instruments were performance-wise.

    ReplyDelete