The words of this tutorial, step by step. The text lives in src/smappy/tutorial/topics/; the directions it follows are in src/smappy/tutorial/STYLE.md.
1. card: Fitting in 2D
What each part of a fitter does, and what to check.
The acquisition is simulated.
spoken: This tutorial goes through fitting step by step, in 2D. 3D has a tutorial of its own.
2. card: What a fitter does
It converts the camera's counts to photons, finds candidate spots, fits a model of the PSF to each, and writes a table of localizations.
spoken: A fitter converts the camera's counts to photons, finds the spots, fits each one, and writes the table.
3. The acquisition goes into source. A Micro-Manager or NDTiff file also tells SMAPpy which camera took it.
4. For a recognised camera these stay on auto: the numbers come from the file and from SMAPpy's camera database.
5. This file has no camera information, so the values are typed in. Camera parameters shows what the fit will use.
6. Every number, and where it came from. A wrong conversion still fits well but gives wrong photon counts, so check it once per camera.
7. Detection smooths each frame and keeps the peaks above a cutoff. The dynamic cutoff is in units of the noise, so it follows the background.
8. Too low a cutoff, and noise is taken for molecules.
9. At the default, the molecules are found and the noise is left alone. Preview is the place to decide this, before a long fit.
10. Gaussian 2D fits the position, photons, background and the width of each spot. Elliptical fits a width in x and one in y.
11. Run.
12. LL is how well the model fits. A spot that is not a single molecule fits badly, and the default filter drops the worst.
13. PSF is the fitted width. Two molecules fitted as one come out too wide; for a 2D fit, the default filter drops them too.
14. Before measuring, correct the drift, with RCC or COMET in the Analysis tab. Filtering and grouping are in the rendering tutorial.
15. card: What to remember
- Camera: check its numbers once.
- Detection: set the cutoff with Preview.
- Model: Gaussian 2D; for 3D, Spline 3D with a bead calibration.
- After the fit: the default filters drop the bad localizations; correct the drift before measuring.
spoken: That is fitting in 2D. Next: fitting in 3D, with a measured PSF.