Including additional variable modifications in the GUI

[gmaddalo79]

Hi,

I wonder if my settings are correct in DIA-NN 1.9.
I wanted to include as variable modification N->D as I did a PNGaseF digestion.
However I do not detect any peptide which such modification, which is a bit strange.

I enlose the report.log file too
report.log.txt

Thanks in advance for your help

[vdemichev]

Hi Gianluca,

--var-mod has one of the two effects in DIA-NN:

• either adding modification to peptides when generating the predicted library or
• scoring localisation of the modification when analysing raw data.
  In your case, if modification is not in the predicted library, it will not be detected.

I would recommend the following steps here:

1. Process in Proteoforms mode in 2.0, apply filtering at PEP <= 0.01. Keep MS2 mass accuracy < 10ppm and MS1 < 7ppm. Ideally as tight as possible until IDs start decreasing.
2. Optionally and if Orbitrap resolution allows: reduce MS1 Mass Accuracy to 5ppm and filter Ms1.Profile.Corr > 0.9.
3. Spontaneous deamidation will still occur anyway, i.e. no need for super-high LC-MS confidence here.

Best,
Vadim

[Obie2020]

Do you know of an R script that extracts modification site abundances (like phosphosites_90/99.tsv or like MaxQuant does for DDA modification site data) from DIA-NN output, which can be tailored to other var mods such as K-GG or K-Ac? So the script would need to name the site at protein level and deal with abundances of several peptidoforms containing the same site (miscleavages, charge states). Related to this: K-GG peptides or peptides with a custom var mod are not yet filtered for any localization probability in the standard dia-nn out, am I right?

[vdemichev]

Replied in another thread. About localisation confidence, it is indeed calculated for all mods declared as --var-mod during search, i.e. also for KGG.