vdemichev / DiaNN

DIA-NN - a universal automated software suite for DIA proteomics data analysis.
Other
278 stars 54 forks source link

Spectral library with modifications #1155

Open cutleraging opened 2 months ago

cutleraging commented 2 months ago

Hello Vadim and Team,

I am using a spectral library for modified histone peptides that I built in skyline. I have some questions regarding using this library...

  1. What is the best way to convert the skyline spectral library (.blib) to a format that will work with DIA-NN? So far I was able to make a .tsv file by following these recommendations: https://skyline.ms/announcements/home/support/thread.view?rowId=65491. I have also attached it for your reference.
  2. When I use this spectra library I encounter this warning: "WARNING: below are the unknown modifications found in the library. While unlikely to have a signficant effect, these might negatively affect identification performance in certain scenarious, it is strongly recommended to declare them using any of --mod, --fixed-mod or --var-mod: +112.1, +126.1, +14, +142.1, +170.1, +226.1, +28, +56, +70, +84.1, +86, +98, +98.1." Since I am looking at histone peptides, there are both fixed and variable modifications. Although this information is all in the spectral library. Is it necessary to specify the modifications as stated in the warning? It appears in my output that the modified peptides were still able to be identified (attached).
  3. When creating spectral libraries, is it always best to use DDA? Or should I use the method that I plan to use for quantification (e.g. slicePASEF).

231123_SC_histones_100cell_spectral_library_DIA-NN_edited.txt

skyline spectral library with 100 cells test.txt

Best, Ronnie

vdemichev commented 1 month ago

Hi Ronnie,

  1. I am not sure if this is possible, but if there are recommendations on Skyline website and they seem to work, please use them.
  2. In this case, it's recommended to declare these modifications (with exact mass, i.e. higher precision than 112.2) using the --mod command, this may improve results.
  3. Can use also DIA or indeed Slice-PASEF. Works fine also with offline fractionation for library creation and in most cases should be more sensitive than DDA.

Best, Vadim

cutleraging commented 1 month ago

Hi Vadim,

Thanks for the response, I will try that out.

That's surprising that a spectral library can be built better with DIA than DDA. Are there any further recommendations for doing this?

Also I need help understanding how the --fixed-mod, --var-mod, and --mod work when using a spectral library. My spectral library has many modified histone peptides. Here is an example: A[+56.026215]AAAAATAPPS[+79.966331]PGPAQPGPR. In this case the +56.026215 is a fixed mod and the +79.966331 is a variable mod. So then I would assume it would be necessary to specify this using --fixed-mod and --var-mod? And what this is then doing is mapping these to the masses, e.g. '+56.026215' would map to 56.026215. But whether I do this or not, it does not seem to really affect the results, which I assume is because of the masses for the fragment in the library which already reflect the presence of these mods. Is my understanding correct here?

Best, Ronnie

vdemichev commented 1 month ago

Hi Ronnie,

That's surprising that a spectral library can be built better with DIA than DDA. Are there any further recommendations for doing this?

Might also make sense to enable peptidoform scoring in DIA-NN when making the lib, will allow to get extra peptidoform confidence. However, in many cases lib-free search will be comparable to a good library in terms of performance.

So then I would assume it would be necessary to specify this using --fixed-mod and --var-mod?

In your case you only need to use --mod. --var-mod enables modification localisation, for example, --fixed-mod is needed for certain scenarious such as multiplexing.

it does not seem to really affect the results It will indeed only have minor effect in most cases.

Best, Vadim

cutleraging commented 1 month ago

Hi Vadim,

Thanks for the replies.

In regards to the modifications, the +56.026215 is actually a fixed mod and like a label. Meaning that it will be on every peptide. So then does this need to be specified in a special way? Otherwise all the other mods are indeed variable mods so then I will specify them using --mod.

Best, Ronnie

vdemichev commented 1 month ago

OK to use --mod for all cases here, including when the modification is ubiquitous. The purpose is to tell DIA-NN the mass of the modification, i.e. [+56.026215] is interpreted DIA-NN just as a sequence of characters, not a mass, hence the need for --mod. The reason DIA-NN does not read masses like that is because often it would be +56 instead of +56.026215, i.e. without the requried resolution, hence more reliable to ask the user to specify all the mods, then try to interpret the names automatically.

cutleraging commented 1 month ago

Okay got it. However, I see that PTMs can also get a Q-value using --ptm-qvalues. Is this relevant here, especially since my spectral library contains many and diverse modifcations?