• Sign In
  1. Requests

Question about plugins and Feature request: Logistic curve fit

support

View Request

view list print
Question about plugins and Feature request: Logistic curve fit MT  2026-03-04 02:05
 
Hello there,

some time ago I asked, if a logistic fit model for calibration curve analysis is available in Skyline. Last time, this was not the case and judging by the changelogs, I believe this is still not the case. Logistic fit models are quite important for absolute quantification of assays utilizing immunoenrichment, as antibody binding kinetics follow a sigmoidal curve. Hence, many calibration curves of IA-LC-MS/MS also describe a slight sigmoidal shape.

I had a look into Skylines source code and I assume, that adding another fit model is doable. However, I am not an expert and don't know the details. Thus I wonder, if it is alternatively possible to write a plugin that does the fitting and backcalculating of values for unknown samples. However I do not understand how well plugins can be integrated to Skyline. Would it be feasible to write a plugin that enables logistic curve fit for Skyline? Will this integrate well into the already established GUIs/workflows?

Best, Tobias
 
 
Nick Shulman responded:  2026-03-04 08:17
The typical thing that you would do if you want to calculate things outside of Skyline is to create a custom report that has all the data that your software will need and then export the report as a .csv file and then run your software on the .csv file.
You can learn about custom reports here:
https://skyline.ms/wiki/home/software/Skyline/page.view?name=tutorial_custom_reports

If you want to go further than that, you could package your tool in a .zip file and upload it to the Skyline tool store:
https://skyline.ms/home/software/Skyline/tools/project-begin.view

The .zip file would have a "tool-inf" folder inside it and some .properties files which tell Skyline what menu items to add to the "Tools" menu.
Additionally, the tool-inf folder could tell Skyline what other software needs to be installed. This software installation really only works for R. It used to also work for Python, but the versions of Python that it works for are much older than what you might be using. We need to overhaul Skyline's support for external tools written in Python.

If you want Skyline to actually display your new type of calibration curve, you would need to change Skyline's source code.

Here is the file which has the regression fit options that Skyline knows about:
https://github.com/ProteoWizard/pwiz/blob/master/pwiz_tools/Skyline/Model/DocSettings/AbsoluteQuantification/RegressionFit.cs
It might be straightforward to tell an AI agent such as Claude Code to implement this logistic regression feature for you.

Here are the instructions for how to build Skyline:
https://skyline.ms/home/software/Skyline/wiki-page.view?name=HowToBuildSkylineTip

If you wanted an AI to help you with building Skyline and implementing this feature you would tell the AI to look at this:
https://github.com/ProteoWizard/pwiz-ai

-- Nick
 
Brian Pratt responded:  2026-03-04 10:30
The important part here is, of course, actually developing the logistic curve fit logic. Approaching the Skyline code can be intimidating for sure, but if you can create it in Python well enough to work off a Skyline report as Nick suggests, it's probably not a big deal for us to implement your logic to Skyline itself. I do think others would find it useful.

Brian Pratt