Should we exclude FCS files (i.e., exploratory PE marker wells) that have too high PE signals in infinityFlow model?

[denvercal1234GitHub]

Hi @ebecht.

Thank you for the package.

You meantioned in #6, "What I think is important is that there should be low spillover from the "exploratory channel" (PE in the BioLegend kit) into backbone channels. Otherwise I believe the models may poorly generalize to other wells for events with high PE intensity."

I can take a look at the SSM spreading matrix for each compensated FCS file, but for the LEGENDScreen experiment, there are a lot of files to do so manually.

QUESTION 1. I have some FCS files that have very very high PE staining (likely due to the fact that the PE-Abs in these wells have non-optimal concentration) and this caused the staining profile of non-PE markers very odd, as discussed in Spectre Issue and shown below: https://github.com/ImmuneDynamics/Spectre/issues/148. Would you advice to not use these FCS files (i.e., these PE exploratory markers) in the infinityFlow models? Or, what is the best practice for this?

An example of a file with very high PE staining and a file with not too high PE staining. You can see that the one with very high PE staining has this diagonal shape of likely technical error because PE-Ab concentration in this well might be too high

QUESTION 2. Is there a way to check whether the spill-over level would still be okay to use the models in R? Like using some sort of thresholds? Or, perhaps, is there a way to check how well/poorly generalizable the models is for a given set of FCS files?

Thank you again.

[ebecht]

Hi @denvercal1234GitHub

I agree with you that there seems to be compensation issues in the file you are showing with "very high PE staining".

For some markers, the PE intensity was saturating the detectors, which caused other channels to be undercompensated (since the true PE intensity was underestimated due to saturation). I don't see a way around this issue so we excluded these events (Sup Figure 15 in the Science Advances paper).

I don't think this is a problem with the amount of spillover itself but rather with saturation of the detectors. I don't know if too high spillover would be problematic, although I suspect the less the better. I haven't studied this so I don't have an answer for how to detect spillover issues besides saturation.

Hopefully this helps, Etienne

[denvercal1234GitHub]

Thank you very much @ebecht!

How did you remove these "undercompensated" events or you remove the whole FCS file that contained these events? Do you have a quick script for this removal you would not mind sharing? I did not see it in the codes deposited.

Thank again for taking the time to respond!