Protein quantification problems

cellprofiler

#1

I have been working extensively with cell profiler for a few months now trying to quantitate protein levels in the nucleus and in the cytoplasm. I all ready have a method of analysis using Metamorph software that I am using as a standard of comparison. When I run my pipeline in the cell profiler software I get very different results from the same images run in metamorph. Upon closer examination it seems the images are somehow being thresholded when they are loaded into cell profiler which looks to be creating the problem. It looks like all the images are being given a best fit, or auto scaled threshold which is giving me fairly equivalent levels of protein expression across images that are clearly expressing different levels to the naked eye and other software.

My questions is: How do I get around this thresholding or is there a way to turn it off and has anyone else experienced this problem?
?

Sincerely,
Sean Burke
Memorial Sloan Kettering Cancer Center
NYC, New York


#2

anyone out there?


#3

there may be a vocation during the laboring days, in May. 1th!


#4

My questions is: How do I get around this thresholding or is there a way to turn it off and has anyone else experienced this problem?

That shouldn’t be the case. What format are the images in? It could be there’s something wrong with conversion from 8/12/16-bit to floating point (CellProfiler converts everything into 0.0-1.0 on loading), which is leading to thresholding at the top end of the spectrum.

If you can send an example image and pipeline to me (thouis@broad.mit.edu), I’ll see what we can figure out.

Ray Jones


#5

Ray,
Thanks for the reply. I just sent an email with all the requested info. Hope to hear from you soon.