-------- Original Message --------
Subject: RE: relative warps to distance matrix?
Date: Fri, 14 Mar 2008 10:41:10 -0700 (PDT)
From: F. James Rohlf <[EMAIL PROTECTED]>
Reply-To: <[EMAIL PROTECTED]>
Organization: Stony Brook University
To: <[email protected]>
References: <[EMAIL PROTECTED]>

Correct, there is no need to weight the relative warps - each is
already multiplied by the square root of its corresponding eigenvalue.
You do need to use a Euclidean distance coefficient based on all of
the relative warps. The result will be identical to Euclidean
distances based on the 'weight matrix' of partial warp and uniform
component scores so the relative warps do not actually need to be
computed for this computation. One can also obtain the identical
distances from the GPA aligned configurations if they have been
projected into the tangent space so the partial warps and uniform
component do not need to be computed either if all you want are these
distances.

The tangent distance described above is an approximation to the
Procrustes distance coefficient (they will be equal to or slightly
smaller). Which you use will depend on exactly what you plan to do
with the distances. However, for most data these two distances are
very highly correlated.

Both distances (and the correlation between them) can be obtained
using the tpsSmall program.

I am not sure what your alternative suggestion is for using the
variance at each landmark.

=========================
F. James Rohlf
Distinguished Professor, Stony Brook University
http://life.bio.sunysb.edu/ee/rohlf

-----Original Message-----
From: morphmet [mailto:[EMAIL PROTECTED]
Sent: Friday, March 14, 2008 12:14 PM
To: morphmet
Subject: relative warps to distance matrix?

-------- Original Message --------
Subject: relative warps to distance matrix?
Date: Fri, 14 Mar 2008 07:09:44 -0700 (PDT)
From: Katja Peijnenburg <[EMAIL PROTECTED]>
To: [email protected]
References: <[EMAIL PROTECTED]>

Dear morphmet,
I have collected shape data of gastropod shells using the TPS
software package and thus obtained 18 Relative Warp scores per
individual. I would like to convert these data into a single
distance
matrix to use in a program 'Barrier' that draws 'barriers' between
populations onto a geographical map. I noticed that the variance of
relative warps is correlated with the proportion of variance they
explain, so there seems no need to weigh the different relative warp
scores. What does the morphmet community think? What would be the
best way to convert Relative Warp scores to a distance matrix, and
what distance should be used?
Or, alternatively, would it be better to use the variance of each
landmark, as was suggested to me by a colleague (using Morpheus)?

Thank you very much for your opinion on this.
Katja Peijnenburg
[EMAIL PROTECTED]


~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Dr. Katja T.C.A. Peijnenburg
Marine Biodiversity, Ecology & Evolution (MARBEE)
UCD School of Biology & Environmental Science
UCD Science Centre West, room 228
University College Dublin
Belfield
Dublin 4
Republic of Ireland
tel. +353.1.716.2013/2262
fax. +353.1.716.1152
http://www.ucd.ie/zoology/marbee/katja_peijnenburg.html
email: [EMAIL PROTECTED]



--
Replies will be sent to the list.
For more information visit http://www.morphometrics.org



--
Replies will be sent to the list.
For more information visit http://www.morphometrics.org

Reply via email to