Monday, May 6, 2013

what to do with this type of data?

A researcher brought this bone marrow sample from a GFP+ mouse.  The unstained GFP+ cells have quite a bit of fluorescence in the 575nm and 610nm channels in addition to the 525nm.  Here are some plots of the data:

First one is singlets from Wild-Type mouse (bone marrow):

Next is singlets from GFP+ mouse (bone marrow):

The bone marrow cells from GFP mice have a lot of fluorescence also in the PE (575nm) and PI (610nm) channels; PI isn't so important for our experiment (thus why the PI voltage probably isn't optimal), but PE is...another marker uses PE as a label:

For comparison, the wild-type cells:

I compensated the FITC and PE channels, thinking this was the optimal/best I could do given the circumstances.  The assumption was that all of the cells are positive, and thus the cells centered around 10^2 are positives. (FlowJo warns me that 6% of the cells are on the chart edges):
I suppose I could compensate MORE?  Maybe the cells centered around 10^2 are negative (different autofluorescence than the wild-type?), and the ones between 10^3-10^4 are the positives?  If I do, now 58% of the cells are on the chart edges!  And I still get a very weird vertical line at the right-hand side.