This post comes from an excellent suggestion for a post in the comments section of something I wrote last week. What about injection times vs. different sample loads and complexities?
For this I'm going to brazenly steal from my absolute favorite talk of 2014 so far. Tara Schroeder of the Thermo NJ demo labs put together at talk for the iORBI tour for obtaining maximum peptide coverage with the QE. I'm going to refer to this slide deck often.
One of the many excellent bits of info is a general starting point for target values and injection times:
Again, this is a starting point. Chromatography conditions will vary a lot, as well the true definition of "simple" and "complex" mixtures depending on what you thing you are looking at (also compared to what you actually are looking at, right? Sometimes they don't exactly line up!)
For something complex and high load (segway, isn't it awesome that we are at a point in time where we are considering over 100 nano grams high load?!?!) we aren't as concerned with hitting our target values are we as about getting as many MS/MS events as possible.
When we drop into the low nanogram range, we are truly concerned that 50ms is not going to be enough time to hit that magic level for each individual peptide that will give us high scores. We sacrifice the number of MS/MS events that we can get in order to increase our chances of getting good ones.
Now, for simple stuff, we simply treat it like low load. By "simple" what we really mean is: that we can easily obtain a single MS/MS event for at least a few peptides for every protein that is present. I think that is a fair starting point. The stress here isn't in getting enough MS/MS events. The real concern is converting every possible MS/MS event into a peptide ID. Again, we sacrifice the number of possible MS/MS events a little in return for giving us twice the possible signal to convert these peptides into high quality MS/MS spectra that the search engine would love.
I want to introduce one other variable here: Dynamic exclusion (previously discussed here.)
For the complex stuff at high load: Almost always, we want to use the soloist approach. 1 MS/MS event and then put the ion onto the exclusion list.
For the complex stuff at low load: This is a toss up. The soloist approach will give you more MS/MS events but at lower efficiency than the two timer approach. Tough to say which will be more effective for a given experiment without more information
For the simple stuff: Two timer! If you've got plenty of cycle time to fragment peptides from every protein present, give yourself a chance to get each peptide at least twice. Yes, the number of unique MS/MS targets decreases (by half), but by claiming your sample is "simple" you've already said that isn't a concern. For single protein pull-downs, I'd allow as many MS/MS events for each peptide as possible. I worked with a group recently where our #1 goal was maximum sequence coverage of a single small protein and its PSMs. Best coverage occurred when we allowed 4 fragmentations of each ion before dynamic exclusion kicked (this gave us a pesky phosphopeptide that just wouldn't ionize well!)
If we go lower, we will need to increase these fill times. But this gives us a crude starting point.
Not sure about your sample complexity or want to double check your run to see if you set it up right?
RAW Meat time!
This is the analysis of the MS2 fill times from an IP run my friend Patricia and I did. The maximum injection time for each MS/MS event was 100ms. We hit the maximum almost every single time. This suggests that we are loading so little sample that we need significantly more than 100ms of fill time. After a pull down it is currently just about impossible to determine your peptide load. We often don't have nearly enough material for a standard protein measurement assay (hopefully someone will come up with something more sensitive soon...) If this were a monoclonal pull down I'd say, crank up that fill time and try running it again!
What do we want to see?
This. This is a low load complex run from my friend Rosa. She used a maximum fill time ~150ms for this run and it was perfectly appropriate for this sample. The first bar represents fill times of <50ms the second represents ~50 ms, the third ~100 and the 4th bar is maxing out. The vast majority of peptides hit target value in less than maximum -- in fact, less half of max fill time. But there were a large number of MS/MS events that required at least half the max and about 1/8 of the peptides needed the full 150ms.
I hope this is clear. Thanks to Kristian for the questions and Tara, Patricia, and Rosa for the data to let me put this together this weekend.
Interesting post, I was just thinking about fill time and ion targets. I was wondering if the Thermo-recommended targets were the minimum number of ions for a good spectra or the max number of ions before space-charge sets in. Based on this it looks like it's more the minimum number of ions for a good spectra.
Yeah, when we recommend fill times for global the unspoken rule is: the minimum amount of time to give us a high likelihood of a peptide spectral match. For simpler or targeted stuff we might start to change our thoughts to the maximum number of ions that we can get without space charging messing it up.
I am #starstruck that you wrote this piece :) Thank you so much! For some reason, none of us in our lab had thought of the "repeat count". I thing that might do a lot for our experiments, and that is why it is awesome that you (as "a top level athlete") write this blog to teach us youngsters the handcraft of mass spectrometry. #BenforpresidentReplyDelete
Lol, thanks for the question, Kristian. I sometimes have writer's block and honestly can't think about what to ramble on about next. Sometimes I just need a good prompt.ReplyDelete
I have been using RawMeat and it works well with Orbitrap XL data, but with QE or QE+, I could not get it to display the MS1 and MS2 fill times. The version I have is 2.1 with MSFileReader 3. What version are you using? -HsinYao
Unfortunately, those functions in RAW meat do not work with Q Exactive or Fusion data. I contacted the author of the software and there are currently no plans to update the freeware.