Re: Re digital data

Actually, there aren't several transforms -- basically there are only 
two ("Logicle", from Dave Parks and Wayne Moore at Stanford, used by 
DiVa and FlowJo; and the "HyperLog" from Bruce Bagwell at Verity used 
by WinList).  And, in fact, these two are mathematically nearly 
identical; the Logicle function is a slightly more complex, and 
slightly smoother version, but probably without noticeable visual 
difference when the same parameters to the function are used.  The 
primary difference between the two is that the Logicle algorithm 
provides for a mechanism to automatically select parameters to the 
function that are optimized to the actual distribution of the 
data--hence, the transformation can, if desired, be stronger or 
weaker for one fluorescence channel than for another (which is 
reasonable, as the magnitude of the error in the measurement is very 
different in these channels, and the error in the measurement is the 
primary reason we need the transforms!).

I am a firm believer in the use of these transformations--in the 
multicolor world, they have revolutionized our ability to look at the 
data.  (Note that the transforms in no way affect statistics or 
gating--they ONLY affect the visual representation of the data).  The 
point of these transforms is to make the data more easily conveyed to 
readers--something that a simple log scale (which is itself a 
transformation!) no longer does adequately.

I would be very interested to hear from anyone who has had trouble 
with reviewers with presenting data like this.	I think in reality 
the reticence is more on the part of the researchers, who are afraid 
of the reviewers.  However, if there are any instances where you have 
encountered resistance from reviewers, please let me know.

The Data Presentation Standards Committee, formed by ISAC, will take 
the responsibility to contact the editors of the Journal with which 
there are problems.  Our goal is to educate not only the reviewers, 
but importantly the Editors--so that they can be made aware that any 
complaints that reviewers might have regarding this visualization are 
baseless.

However, my guess is that there have been no problems with 
reviewers--hence, I would like to stem the tide of urban mythology, 
and prevent this from becoming a virtual problem!

mr


>Dear all , in reply to Jim's email
>		   I too am puzzled by the lack of digital data that 
>has been published. In contrast to Jim however, I find that the 
>majority
>
>of my users want to use the "log  / biexponential " transform.They 
>find , and I agree, that the compensation is more intuitive and that 
>all the
>
>data can be seen. Questions that occur to me , which transform ? as 
>there 're several and are they all equally valid.
>It would be useful, to say the least, if there were some consensus 
>on this , between those who review the data, software suppliers, the 
>people
>
>who set the standards for such things , instrument manufactures and 
>the originators of these functions( transforms) !
>I think it extraordinary that this issue is so muddled. Do we wait 
>until investigators have papers rejected on the basis that the data 
>representation
>
>is unacceptable. I think this issue is important and needs to be 
>addressed so we have a clear idea of the usefulness of this 
>representation.
>
>thanks
>
>		    ralph
>
>Ralph Rossi
>Flow facility Manager
>Peter MacCallum Cancer Institute
>Melbourne, Australia
>email	r.rossi@pmci.unimelb.edu.au