Fix for "A duplicate key has already been added" exception during chromatogram extraction…

[bspratt]

… when two molecules in different molecule lists have identical precursor but different fragments.

Reported by user Anna Illiano.

[nickshulman]

Currently, the difference between "AreEquivalentGroupNodes" and "AreEquivalentGroups" is that the "AreEquivalentGroupNodes" also calls "EquivalentChildren".

With this change, "AreEquivalentGroups" also calls "EquivalentChildren" if they're non-proteomic.

I wonder if this means that we should combine "AreEquivalentGroupNodes" and "AreEquivalentGroups". I do not understand when someone would need to call one or the other.

(I don't understand this code very well at all-- this is just something I noticed)

[brendanx67]

Currently, the difference between "AreEquivalentGroupNodes" and "AreEquivalentGroups" is that the "AreEquivalentGroupNodes" also calls "EquivalentChildren".

With this change, "AreEquivalentGroups" also calls "EquivalentChildren" if they're non-proteomic.

I wonder if this means that we should combine "AreEquivalentGroupNodes" and "AreEquivalentGroups". I do not understand when someone would need to call one or the other.

(I don't understand this code very well at all-- this is just something I noticed)

[brendanx67]

Oops

[brendanx67]

This looks like a really odd change to me. The original AreEquivalentGroups function appears to be targeted at deciding whether to precursors are equivalent within the same molecule/peptide, i.e. they share an adduct and label type. Those are the only two distinguishing features between two precursors of the same molecule.

I am really not clear on why this would cause chromatogram extraction to throw an exception. But, perhaps the function is getting misused somehow or overly overloaded from its original purpose.

[bspratt]

I guess this is old enough SRM handling code that none of us wants to claim responsibility for it. But if you run the updated test without the changed AreEquivalentGroups you can quickly arrive at the problem, which as I understand it is that two groups are being "merged" even though they have dissimilar children. Near as I can tell that was never a problem with peptides since the children are generated and can't help but be the same.

[nickshulman]

I believe this recent support request also touches on some of the same issues:
https://skyline.ms/announcements/home/support/thread.view?rowId=52220

Things in the .skyd file are identified by the peptide/molecule text id plus the precursor m/z.
If two different precursors in the Document happen to have the same key, then their chromatograms get stored in the same place in the .skyd file, and weird things may result.

I wonder if we should add stuff to the .skyd file so that two different peptides or molecules from the document are guaranteed to have separate keys in the .skdy file.

[bspratt]

Closing this, Nick is going to take a different approach to the problem