Determine FD baseline interactively about pyjibe HOT 6 CLOSED

As all preprocessing steps are currently non-interactive, this would mean that the user interface would have to be redesigned (which is possible of course).

The corresponding function (correct_tip_offset) is implemented here:

• https://github.com/AFM-analysis/nanite/blob/master/nanite/preproc.py#L87-L94
• https://github.com/AFM-analysis/nanite/blob/3cc367872afd67b639dbd8604b46b82e4cea8f8c/nanite/indent.py#L151-L212

As you can see, this is already a little sophisticated.

I know that this is a design flaw in PyJibe. But still, would a workaround also work for you?
E.g. a preprocessing step that

• cuts 25% of the approach segment or
• corrects for a linear tilt in the approach segment?

Maybe you can attach a screenshot or sample data so that I know what I am up against.

from pyjibe.

I am attaching here two images, one having a baseline that was corrected by PyJibe (1st image) and one where I entered manually the baseline value before the contact point. What I think could be a good solution is to take the values a bit before the contact point, would that be possible or the fitting should happen first? what I also noticed that I cannot really make the fitting range limited so I don't have to fit also the tilted area, even if I change the fitting range on the top it needs a new baseline value.
I attached also the data of the curve I am using in the images
Cell1-2019.02.18-11.24.27.797.jpk-force.zip

from pyjibe.

Since the preprocessing takes place before the fitting and the contact point is a result of the fit, it is not possible to estimate the force baseline from values before the contact point.

I think a solution could be to add a linear model with negative slope to the fit. In the UI, this would then become just another checkbox like "Suppress residuals near contact point". For the example you showed, this should give you solid results. How does that sound?

The question remains how to interpret a tilted baseline.

BTW I think you attached a different file.

from pyjibe.

I think that usually the inaccuracies in the baseline don't always come from a tilt and that is why I am not sure we should fit the linear model with the negative slope and subtract it from all the points. I have few curves that show some weird up and down at the beginning (something based above the cantilever before contact). or some that have oscillatory noise at the beginning. For now I can type the baseline manually but would it be possible to fix the bug with the fitting range? when I try to change it PyJibe crashes. I can then limit the fitting to be around the contact the point.
I checked the file and it is the same name as the one plotted in the screen shot, but maybe it look different because it is zoomed on?

from pyjibe.

I was suggesting to remove the tilt only from the values above the contact point. If we did that and additionally add a preprocessing step that removes let's say 25% of the approach curve, would that work for you? Maybe you can upload some of the curves that exhibit the noise you described.

The fix for the fitting range (b489d28) will be in the next release.

The file was correct - sorry, my mistake.

from pyjibe.

OK, closing this one according to what we discussed in our meeting (tilt not physical, cut not necessary).

from pyjibe.