Dear Skyline team,
Hope you're doing well. I'm excited by the new feature finding feature but have some hopefully helpful feedback regarding its use. Specifically, I work with negative mode small molecule data, so this may just be an unintended use case.
Skyline assumes the feature finding will return protonated molecules, so it returns the chemical formulae with M+H adducts, despite Hardklor feature finding being performed on a negative mode only raw file. This then causes an issue when it tries to import the raw file as only positive mode transitions are present in the transition list. When I manually modify the adducts and chemical formula to reflect negative mode, the correct molecules are observed and extracted. So Hardklor/Bullseye might be working fine, and only changes to the transition list processing may be needed.
If small molecule use is unintended, it might be helpful to exclude it from the small molecule mode view.
Happy to provide all the associated files if desired.
Cheers,
Chris
|
| |
| Brian Pratt responded: |
2024-05-18 00:08 |
Hi Chris,
Hardklor as written does assume positive mode peptide data, I should add a check for negative mode data so as not to waste anyone’s time, sorry.
Though it’s interesting to think about possible changes - Hardklor does permit custom Averagine models so conceivably Skyline could be extended to support your use case of non-peptide molecules.
Thanks for the feedback!
Brian Pratt
|
| |
| Chris Ashwood responded: |
2024-05-26 17:32 |
Hi Brian,
Depending on how much the team wants to support non-routine use, negative mode only needed a few tweaks to get it working well, though I could see issues appearing if a single run contains polarity switching between both positive and negative mode.
No problems! Always happy to try out the latest features, thanks for working on them.
Cheers,
Chris
|
| |
| Brian Pratt responded: |
2024-05-27 10:39 |
I'm intrigued - what were the tweaks?
I'm concerned, though, that there may be built in biases in Hardklor that assume behavior (waving my hands here) specific to protonation - I know that deprotonation results in rather different peptide fragmentation, for example, though that's not a great example since we're only examining MS1. Just a vague sense that the results could be misleading - perhaps others on this list have ideas.
Polarity switching would certainly complicate things, yes. But that's a less common use case, thankfully.
Brian
|
| |
| Chris Ashwood responded: |
2024-05-31 05:51 |
Hi Brian,
There were two required tweaks:
- Hardklor writes the adduct as M+XH, whereas negative mode is M-XH. So I replaced the + with a -.
- The chemical formulae that Hardklor reports are based on the calculated masses which involve the assumption of protonation. As a result, the chemical formula has too many hydrogens, corresponding to the assumed protonation charge state. So I needed to change the number of hydrogens on the chemical formula for Skyline to extract the exact same m/z value with the right adduct.
E.g. we have a M-2H glycan that is 1111.3955 m/z (observed). Hardklor assumes that it is M+2H, giving a neutral mass of 2218.7607 which then affects the chemical formula determination. To then extract out 1111.3955 m/z from the data that went into Hardklor (in Skyline), we need to assign it the M-2H adduct and adjust the chemical formula to match that difference. So we add 4 hydrogens to account for that difference.
Cheers,
Chris
|
| |
| Brian Pratt responded: |
2024-05-31 12:12 |
So you were altering the Hardklor source code?
|
| |
| Chris Ashwood responded: |
2024-05-31 13:20 |
No, just the output transition list itself.
I would export it from Skyline, modify the molecular formula and adduct columns, and then reimport the raw files.
|
| |
|
|