Quanto user guideAll documentation

SCAN deconvolution

Deconvolution splits a full-scan chromatogram into components, each with a spectrum and a reconstructed trace. Compound Discovery runs it when the source is SCAN. SIM and MRM discovery use the acquired channels instead of this scan walk. See Compound Discovery.

What it does

In plain language:

  1. Every scan is read as a list of masses and intensities.
  2. Masses are grouped into features. A feature is one m/z tracked across scans, within 0.2 amu.
  3. Around each feature, a short chromatogram is extracted (0.2 minutes before and after the seed).
  4. That trace is Savitzky–Golay smoothed (window 9, order 5) and integrated. The peak nearest the seed is kept.
  5. Apex times are shifted to account for when that mass was scanned in the quadrupole (high m/z first).
  6. Peaks whose apex times fall within the Corr. window (default 0.02 minutes) are clustered into one component.
  7. Ions shared by overlapping neighbors are split, unless the front or the tail is a normal chromatographic edge.
  8. Extra peaks on a reconstructed trace become their own components only when a valley separates them.
  9. Components closer than 0.02 minutes with a co-elution score above 70 are merged. That merge uses those fixed engine numbers. It does not read the Co-elution box.
  10. Ions below the intensity floor are removed. The floor is the larger of Min ion intensity and Min spec peak % of the component height.
  11. Each remaining component gets a spectrum and a reconstructed chromatogram. The spectrum sent to library search is marked Deconvoluted.

Independent masses are processed in parallel. The progress overlay reports the stage (building the pool, extracting peaks, clustering, and so on) and a percent. Cancel stops the run.

Deconvolute uses From and To. Deconvolute all clears the time limit and runs the whole TIC. A missing TIC span reads every scan.

Settings you can edit

These are on the chromatogram header. They are saved when you edit them. They apply on the next Deconvolute.

Setting Default Effect on the next SCAN run
Corr. window 0.02 min Peaks whose apex times differ by less than this are one component. If you set it to 0, the engine uses 0.01 min.
Co-elution 0.60 Saved with the view. The SCAN run does not read this number. Merging of nested components uses the fixed rule in the steps above. SIM and MRM grouping uses Corr. window (or 0.05 min when that window is 0).
Min spec peak % 0.5 Percent of the component height. Masses below this are left out of the spectrum. The spin stops at 50. The larger of this floor and Min ion intensity wins.
Min ion intensity 1 Absolute intensity. Masses below this are left out after the component is built.

Settings inside the engine

You do not edit these on the header. The SCAN run uses them as follows:

Setting Value used Effect
Minimum signal-to-noise 3 Features quieter than this are not kept.
Minimum ions in a component 1 A component with no ions left after the intensity floor is dropped.
m/z range 1 to 1 000 000 The SCAN adapter does not apply a 35–800 GC window.
Mass tolerance 0.2 amu How far from the seed mass an ion may be and still belong to the trace.
Extract half-window 0.2 min Chromatogram extracted around the seed.
Savitzky–Golay window 9, order 5 Smoothing before the feature is integrated.

There is no separate “sensitivity” or “resolution” knob. A larger Corr. window merges peaks that apex further apart. A higher Min ion intensity or Min spec peak % drops more noise and can drop real low ions.

The component spectrum is searched as origin Deconvoluted. Prescreen and ranking then treat it as a cleaned spectrum (forward match, stricter unexplained peaks than a raw apex). See Library search.

Long chromatograms are slower because every scan and every mass is visited. Cancel from the progress overlay if you only needed a short From / To range.