Re: Contour plots & smoothi

Alan Stall (astall@pharmingen.com)
25 Sep 1997 16:04:54 -0700

RE>>Contour plots & smoothing: rights... 9/25/97

I'll add my two-cents worth.

Dot plots have one, count 'em one, major advantage. They are very good at
displaying small (<2%) and/or diffuse populations such as single positives in
the thymus or CD5+ B cells in the spleen.

Dot plots have two MAJOR disadvantages. First they are very bad for visually
estimating relative population densities for populations which both constitute
major populations. Furthermore in order to get a visually informative display
you have to play around with how many dots are displayed in the plot. If you
are looking at a 0.1% population and are displaying 100K events in order to
cleanly reveal your population, the rest of the plot is going to look blacker
than a cave at midnight.

The second problem is that dot plots are very poor at resolving two
populations; such as a small pop. near a large pop. or two pops. with two fold
mean differences. Again you can play with the number of dots to help it look
better but it's a waste of time. You just can't really do it effectively.

On the other hand contour plots (especially probability plots) are extremely
good at the latter two functions. With a correctly calculated prob. contour
plot you can readily identify a "shoulder" pop. which constitutes only 10% of
the major pop. In addition the presence of two populations whose MFIs differ
by only 50% are also easily identify. I will admit that the "learning curve"
on reading contour plots is a little steeper. It's like learning to read
elevation maps. You have to distinguish the hills from the valleys. With all
my students there was an initial frustration followed by an "AHA" light bulb
realization. After that point they would never use dot plots unless they were
looking at small populations (and even then log contour plots were sometimes
more infomative- but that's another discussion).

So it's not just a question of which way is better. It depends on the type of
data you are presenting. With most of the software on the market you have to
make a choice- either dot or contour. In that case, if you are looking at
small populations use dots in all other cases I would recommend using contour
plots.

The ideal situation would be to have a type of plot that has the advantages of
both dot plots AND contour plots with none of the disadvantages. That is
exactly what a contour plot with outliers is. A contour plot in which any
populations which lay outside of the last contour are displayed as dots. The
best of both worlds! I think that this is the plot type of the future. To my
knowledge the only software currently available which has this option is
FlowJo. I am sure other software programs will follow.

Alan Stall

--------------------------------------
Date: 9/25/97 3:09 PM
To: cyto-inbox
From: Kenneth A Schafer

Flow-ers,

Mario Roederer said:

> Of course, you will agree that dot plots are completely inappropriate.
> (Everyone: please stop publishing data with single-color dot plots!)

You've lost me on this one. Being a relative new-comer to flow
cytometry, I don't see what is wrong with dot plots. Please educate
me. If dot plots are "completely inappropriate", is there ever a
situation when they would be appropriate to use?

Thanks,
Ken Schafer
kschafer@lucy.tli.org
Kenneth A Schafer
Lovelace Respiratory Research Institute
PO Box 5890
Albuquerque, NM 87185
505-845-1126
505-845-1198 (fax)


------------------ RFC822 Header Follows ------------------
Received: by pharmingen.com with ADMIN;25 Sep 1997 15:08:00 -0700
Received: from flowcyt.cyto.purdue.edu ([128.210.60.31]) by fw.pmgsd.com
via smtpd (for [204.182.230.3]) with SMTP; 25 Sep 1997 22:05:10 UT
Received: by flowcyt.cyto.purdue.edu (940816.SGI.8.6.9/930416.SGI.AUTO)
for cyto-sendout id OAA07063; Thu, 25 Sep 1997 14:15:46 -0500
Received: from audrey.tli.org by flowcyt.cyto.purdue.edu via SMTP
(940816.SGI.8.6.9/930416.SGI.AUTO)
for <cytometry@flowcyt.cyto.purdue.edu> id KAA06052; Thu, 25 Sep 1997
10:16:29 -0500
Received: from itri-1.lrri.org ([204.134.38.201]) by audrey.tli.org
(4.1/SMI-4.1)
id AA29473; Thu, 25 Sep 97 09:00:01 MDT
Received: from ITRI-1/MERCURY by itri-1.lrri.org (Mercury 1.21);
25 Sep 97 08:57:15 MDT
Received: from MERCURY by ITRI-1 (Mercury 1.21); 25 Sep 97 08:57:06 MDT
From: "Kenneth A Schafer" <kschafer@itri-1.lrri.org>
Organization: LRRI
To: cyto-inbox
Date: Thu, 25 Sep 1997 08:57:01 MDT
Mime-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7BIT
Subject: Re: Contour plots & smoothing: rights and wrongs
Priority: normal
X-Mailer: Pegasus Mail for Windows (v2.42a)
Message-Id: <2262E3B7F@itri-1.lrri.org>