PTM likelihoods and scores

[acesnik]

I'm reading this MoFi paper right now, and it's reminding me of @veitveit's concern that we need to allow room for scores in PTMs.

I think we might need to formalize a solution for at least two categories of scores:

1. likelihoods of localization

• Perhaps PROT[Phospho|#mod:20]EOFORMS[Phospho|#mod:80], as noted in Ambiguity of PTM localization #17

2. likelihood of different modifications at the same position accounting for the same mass difference

• We have come upon this problem in Proteoform Suite, and they've come upon it in this paper.
• In the text they give the example (with glycans): "For instance, assume that a given residual mass may be compatible with the glycoforms A2G0F/A2G2F and A2G1F/A2G1F, whose scores are 0.7 and 0.3, respectively. Then, the former permutation will account for 70% of the peak abundance, while the latter one will explain the remaining 30%."
• How should we note this type of likelihood for two different modifications at the same position?

Can anyone think of more categories of scores?

[acesnik]

In an attempt at 2), I suppose the ambiguity grouping works nicely even for the same position:
PRO[A2G0F|#glycan:70][A2G1F|#glycan:30]TEOFORM

And with two positions, that is also possible, where the probability for the glycan is 70/30 at each site and the position likelihood is 80/20:
PRON[A2G0F|#glycanA:70|#position:40][A2G1F|#glycanA:30|#position:40]TEOFORN[A2G0F|#glycanB:70|#position:10][A2G1F|#glycanB:30|#position:10]

[acesnik]

I'm actually confused by my own follow-up statement from last week. The # mark would usually be used for the same modification at several locations, indicating it could be at one of the set. The #position mark breaks that assumption, since it is used with a couple types of modifications, as does #glycanA and #glycanB.