Questions about RANSAC algorithm

[grimace9527]

Is your feature request related to a problem? Please describe.
Dear mzmine team,
Thanks for your novel and user-friendly software. And this is actually not a feature request.
Since I'm new to quantitation of proteomics, when I am trying to learn the RANSAC algorithm in the MZmine2 article (https://bmcbioinformatics.biomedcentral.com/articles/10.1186/1471-2105-11-395), some questions are encountered. Could you please explain the algorithm briefly to me and thanks a lot for your reply!

About the sentence "The purpose of peak list alignment is to match relevant peaks across multiple samples.", I am curious about how the "relevant peaks across multiple samples" are decided or calculated. Are these peaks actually the "features" extracted from MS1 spectra? If yes, how to select the features used to align? If no, then what kind of peaks are selected?
And if the data used for alignment are just 1-d arrays containing time points?

Thank you again for your kind reply. : )

Best Wishes,
Jim Thompson

Describe the solution you'd like
A clear and concise description of what you want to happen.

Describe alternatives you've considered
A clear and concise description of any alternative solutions or features you've considered.

Additional context
Add any other context or screenshots about the feature request here.

[tomas-pluskal]

Hi Jim,

The principle of the algorithm is described in the same text, just below the part that you highlighted. The algorithm uses the m/z and the RT of each peak (feature) for the alignment. If you need more details, you can check the code here:
https://github.com/mzmine/mzmine3/blob/0b4186061b0f04cf8b04459b336bf4852cad58bf/src/main/java/io/github/mzmine/modules/dataprocessing/align_ransac/RansacAlignerTask.java#L127

Best regards,

Tomas

[grimace9527]

Hi Dr. Tomas,
Thanks for your response and I think I get the principle the algorithm now. But I have just one more question...If all the features which passes the filteration of the "RANSAC aligner" window are used for the alignment, is there a set of preferred parameters based on your experience? Thanks a lot!

Best regards & thanks
Jim

[tomas-pluskal]

Hi Jim, I personally use the Join aligner. because it is conceptually more simple.
You should try to optimize the parameter values based on your dataset.

[grimace9527]

Thanks for all your response! : ) I'll try the optimization