----->>>>> On 17-Mar-1999, Ray Hicks wrote: > The reason for deconvolution is to create a perfect world > where those assignments can be made. The conspiring spread > functions are modelled and how well the model's output > matches the real histogram is a measure of perfection. But > the nature of modelling still doesn't allow you to identify a > cell. > > Jill wanted exact boundaries that she could gate on, I don't > see how modelling could give them. It identify a point where > there is a satisfactory tradeoff between contamination from > an unwanted compartment and loss of the required one, but > would it improve the analysis of her horseshoes where she > already has an indicator of S-phase? -------------- Doesn't this argument come down to a differentiation between a sort gate and an analysis gate? At collection time, you don't want to be making decisions about an individual cell, but in post hoc analysis, there is potentially interesting information to be gleaned from looking at the overlap population, or from subtracting it out. We've been debating this issue internally. I think it makes sense statistically, and am lobbying for it in the next FlowJo. Mario is the self imposed guardian of scientific rigor and likes to hold his breath until he turns blue. But for me, the role of data analysis software is to enable exploration and to test potential models, not only to churn existing models. Mark's message lists valid uses for the capability. Clearly there are limitations, but this is flow, so what else is new? The software is there only to calculate the numbers; it's up to the biologist to make sense of them. Adam ----------------------------------- Adam Treister adam@treestar.com 650-508-9349 http://www.treestar.com -----------------------------------
This archive was generated by hypermail 2b29 : Wed Apr 03 2002 - 11:53:16 EST