Infected cells are "always" brighter in serum than non-infected

hci-unihd / antibodies-analysis-issues

Issue tracker for problems in the antibodies analysis workflow.

0 stars 0 forks source link

Infected cells are "always" brighter in serum than non-infected #33

Open tischi opened 4 years ago

tischi commented 4 years ago

@Steffen-Wolf @metavibor @imagirom

Is it still the case that infected cells are always a bit brighter in the serum channel than non-infected? I made one scatter plot for one plate where it seems to be the case. I don't know if that changed in the newer plates...

Anyway, I could think of two reasons

Maybe there is always some (unspecific) immune response in all people
Maybe there is some fluorescence crosstalk from the virus channel into the serum channel?

For the second point: @metavibor did you already image a plate where you leave out the serum staining but otherwise acquire it with the same settings as usual?

metavibor commented 4 years ago

Hi Tischi, Yes that is one of the things I was wondering as well. I was suspecting your "point no. 2" might be the reason but than I stumbled upon a well where there was no serum labelling but the ratio was 1.1 or something like that. I think I even reported that in one of my emails... I will look again which well was that exactly. I also noticed that this seems not to be the problem any more. For example the analysis of the two last plates "316" and "764" where I had two well without labeling gave ratio of 1.008... this "008" might be within the error or it could be attributed to one of your points. Also all other negative control gave ratios of 1.0XX... which is great. In any case I was planning to take some proper crosstalk evaluation once I zero in on exact acquisition settings so that we can get rid off as many imperfections as possible,

tischi commented 4 years ago

I was suspecting your "point no. 2" might be the reason but than I stumbled upon a well where there was no serum labelling but the ratio was 1.1 or something like that.

I don't understand your "but" in above sentence. This is exactly what I would expect for a fluorescence crosstalk issue: a ratio of 1.1 in the absence of serum staining, isn't it?

Anyway, for the classification I think it is not such a big deal, because we can just slightly adapt the threshold, but, e.g., for a publication I think it would be good to be exactly sure what is going on.

metavibor commented 4 years ago

sorry, forgot to mention, there was also no "virus" labelling, there was only DAPI in that well... so I could not understand where the crosstalk would come from. Only weak cellular background was present in the "virus" channel...Now on the second thought, there could have been some dirt in the TRITC channel that was recognized as "virus labeling" and the crosstalk would make sense...

tischi commented 4 years ago

...at least in below plate there still is an offset. Not sure this is already one of the "new ones"?

and here is another example:

metavibor commented 4 years ago

indeed... the second example is from the new plate where I reduced "virus labelling" by a factor 5 compared to earlier plates and it looks slightly better than your first example. But the offset is still there. I think it is worth doing crosstalk assessment on the next plate to see if we can get rid of this.