MannLabs / alphaviz

AlphaViz is a cutting-edge browser-based interactive visualization tool allowing to visualize the processed mass spectrometry data acquired with Bruker instrument.
Apache License 2.0
30 stars 2 forks source link

Intensity calculation for MS2 #10

Open straussmaximilian opened 2 years ago

straussmaximilian commented 2 years ago

Hi, I did some more testing and wanted to investigate some AlphaPept vs MaxQuant results in AlphaViz. Here I noticed that the intensity values for MS2 fragments are different. More specifically, the peaks don't seem to be that intense. Is this expected? In AlphaPept I use .readPasefMsMs(), so maybe @swillems has a clue where this could be coming from.

For reference, this is the spectrum I see in AlphaViz(one-click installer v.1.07) Screen Shot 2022-02-25 at 11 17 22 AM Below the spectrum I get from an AP search: Screen Shot 2022-02-25 at 11 18 34 AM The axis description seems to be correct, and the fragment intensities are low. Screen Shot 2022-02-25 at 11 19 39 AM

On another note: Clicking on Autoscale Screen Shot 2022-02-25 at 1 42 03 PM seems to squeeze the sequence annotation of matched ions. Screen Shot 2022-02-25 at 11 19 15 AM

Additional Feature Suggestion: Right now I need to click the protein to get to the peptide I want to investigate. Here it would be cool if I could directly look up a sequence w/o knowing the protein.

swillems commented 2 years ago

The readPasefMsMs returns centroided spectra. I assume AlphaViz shows the actual raw data which is in profile mode. Could you try to zoom in on on of the AlphaViz peaks with a high intensity and check that these indeed have a larger area under the curve which will be summed after centroiding?

straussmaximilian commented 2 years ago

Screen Shot 2022-02-25 at 2 54 30 PM Screen Shot 2022-02-25 at 2 54 49 PM

It seems indeed this is profile data. However, then I find the current annotation confusing. The Error plot will give an error for a centroided peak, and the color annotation is also for a single peak. Suggestions: One could plot profile spectra as line plots to avoid confusion. For the color highlighting, one could color all peaks that are within the selected tolerance and not just the closest.