Small Molecule MRM extraction with isomers

support
Small Molecule MRM extraction with isomers wbarshop  2018-08-13
 

Hello Skyline team,

Briefly:
I have run into an issue where transitions which are acquired in our method are not properly being associated to the small molecule inside Skyline. This results in "empty" transition data.

More long-winded:
I have been working on a project where we are quantifying small molecules which exist as multiple isoforms and are separable chromatographically.

We optimized collision energies via direct infusion, and started by utilizing the observable fragments and their empirical masses. Of course, when we switch from isoform to isoform for this process on our QQQ, the masses reported back vary a little, even if the fragments being monitored are the same.

My suspicion [based on very little] is that when we have multiple small molecules where transitions exist with the same (or very similar) m/z values for both the precursors and fragments, the matching behavior of Skyline (via the "Method match tolerance m/z" setting) may be getting tripped up and leaving out data.

I have included a simplified Skyline document with four analytes. All share the same precursor m/z, several of them share all transitions (often within the limit of the method match tolerance m/z setting). In this case, you can see the "Deoxycholic acid" analyte loses its transition extraction (despite skyline having the exact m/z values stated by the instrument method-- though I know some values like monoisotopic m/z and average m/z seem to be float representations and not precise).

In this Skyline file, you'll notice that Deoxycholic acid fails to extract its fragment ions. If you modify the small molecule and remove the explicit retention time, you'll see that we extract the peaks from the "3b12a" molecule, as they are within the method match tolerance.

Even, though, if I drop the method match tolerance to be very narrow, I cannot seem to recover the extraction of the proper MRM data for "Deoxycholic Acid."

I am included here the simplified skyline file, along with the same RAW file which is imported into the sky file. Please let me know if you need any additional information or data, I will be happy to provide.

Best,
William

 
 
Nick Shulman responded:  2018-08-13
I think you attached a slightly different version of the Skyline document than you meant to.
The explicit retention time for "Deoxycholic acid 3a12a" is blank and it has a full set of chromatograms.
What did you have the explicit retention time on that molecule set to which caused its chromatograms to disappear?

When you import results from an SRM file, all of the chromatograms get copied into the .skyd file. Usually this means that if you make a change to something like Explicit Retention Time, Skyline is immediately able to change which chromatograms are assigned to that molecule. However, if you have two molecules which, at the time of import, Skyline decides should use the same set of chromatograms, then Skyline puts two copies of the chromatograms into the .skyd file, and each copy explicitly remembers which molecule it belongs to. In this case, after you make a change to something like "Method Match Tolerance m/z" or "Explicit Retention Time", you probably also need to do "Edit > Manage Results > Reimport" in order to see the full effect of that change.

Let me know what the explicit retention time of that molecule is supposed to be, and I will try to figure out what is going on.
 
wbarshop responded:  2018-08-13
Nick,
Thanks for lending a hand!

Sorry! You're absolutely right, I uploaded a copy of the skyline document while I was in the middle of tweaking it.
I also recognized an error in my old file where the quad resolutions were improperly set to be larger than necessary. In changing some of these parameters, the behavior has slightly changed, but the issue of failure-to-associate the data with deoxycholic acid persists.

The explicit retention time for Deoxycholic should be 10.47min (as it exists in the acquisition method).

If you leave the method match value to .1 (which should be fine as the transitions all match to acqusition method well within this tolerance) with explicit retention time on, it seems to miss the fragmentation data entirely.
If you take away explicit retention time and leave method match to .1, it will still miss all transitions. And once you expand the method match value up higher (like, to .4 or .6) then we will get the transitions from the scheduled acquisition of the other analytes -- but NOT the data for deoxycholic acid.

I have reuploaded the shareable skyline file, and a "sanity check" screenshot -- a view of the data in QualBrowser showing that this raw file does, indeed, contain data for deoxycholic acid at the retention times around 10.47min.

I am including the small molecule transition list import table below, in case that is of use in looking through this small file or recreating it with small tweaks:

Molecule List Name    Precursor Name    Precursor m/z    Precursor Charge    Explicit Retention Time    Explicit Collision Energy    Product m/z    Product Charge
Deoxycholic acid 3a12a    Deoxycholic acid 3a12a    391.282    -1    10.47    33    327.28    -1
Deoxycholic acid 3a12a    Deoxycholic acid 3a12a    391.282    -1    10.47    38    343.294    -1
Deoxycholic acid 3a12a    Deoxycholic acid 3a12a    391.282    -1    10.47    33    345.276    -1
Deoxycholic acid 3a12a    Deoxycholic acid 3a12a    391.282    -1    10.47    10    391.282    -1
Hyodeoxycholic acid 3a6a    Hyodeoxycholic acid 3a6a    391.283    -1    8.36    45    347.296    -1
Hyodeoxycholic acid 3a6a    Hyodeoxycholic acid 3a6a    391.283    -1    8.36    35    355.264    -1
Hyodeoxycholic acid 3a6a    Hyodeoxycholic acid 3a6a    391.283    -1    8.36    30    373.275    -1
Hyodeoxycholic acid 3a6a    Hyodeoxycholic acid 3a6a    391.283    -1    8.36    10    391.283    -1
3b12a    3b12a    391.3    -1    8.85    35    327.4    -1
3b12a    3b12a    391.3    -1    8.85    35    343.4    -1
3b12a    3b12a    391.3    -1    8.85    30    345.4    -1
3b12a    3b12a    391.3    -1    8.85    10    391.3    -1
Chenodeoxycholic acid 3a7a    Chenodeoxycholic acid 3a7a    391.3    -1    10.21    30    329.284    -1
Chenodeoxycholic acid 3a7a    Chenodeoxycholic acid 3a7a    391.3    -1    10.21    30    347.295    -1
Chenodeoxycholic acid 3a7a    Chenodeoxycholic acid 3a7a    391.3    -1    10.21    35    373.274    -1
Chenodeoxycholic acid 3a7a    Chenodeoxycholic acid 3a7a    391.3    -1    10.21    15    391.3    -1



Let me know if there are any other issues or questions!

Cheers,
Will
 
Nick Shulman responded:  2018-08-14
Skyline seems to have a bug in it.

Your molecule has a precursor 391.282 and product ions 391.282, 345.376, 343.394, 327.282.
The best match ought to be the chromatograms in your .raw file: 391.282 & (391.282, 345.278, 343.294, 327.28).
However, Skyline decides that it is a better match to 391.3 & (391.3x2, 345.4, 343.4, 327.4).
There are a couple of reasons that Skyline is doing this incorrectly:
1. Skyline is double-counting the 391.3/391.3 match to your precursor, because you have two identical chromatograms with that precursor/product in the .raw file.
2. Skyline is using the maximum possible method match tolerance (0.6) instead of whatever tolerance you have specified in your instrument settings.

I will try to fix this.
 
wbarshop responded:  2018-10-02
Nick,

Sorry to bump this back up. Is my best course of action for the time being to drop the precursors from chromatogram extraction, or do you suppose that this multi-matching problem will be present if two (or several) analytes have identical product ions?
Alternatively, would I be better served by introducing a small "wiggle" in the actual fragment masses so that no two analytes have the same fragment ion mass (eg 391.3 & (391.2820) and another compound 391.3 & (391.2821)) paired with shrinking down the method match tolerance to 0.0001?

I tested for this problem on the most recent daily release to find it persists.


Best,
Will
 
Nick Shulman responded:  2018-10-02
Oops. Sorry I forgot to fix this one. Let me get back to you in a day or two.
Hopefully this is straightforward to fix.
I actually cannot think of any decent workarounds.
 
Nick Shulman responded:  2018-10-08
This bug will be fixed in the next update of Skyline-Daily.

It is also already fixed in the latest Skyline Daily Administrator Installer (4.1.9.18281), which you can install from here:
https://skyline.ms/wiki/home/software/Skyline/daily/page.view?name=install-administrator-64