Re: [MORPHMET] Appending tps curves to landmarks

[Liu Idarraga Orozco]

I had de same problem in times several. If I remember correctly the 
problem comes out when some comma is in the scale number or something 
like that.


Liu


On 10/11/17 08:25, Candice Neves wrote:

Hi All

I'm currently trying to append curves to landmarks using the function 
in tpsUtil (I want to use the curve information as sliding landmarks) 
however any time I try to do this tpsUtil gives me an error message 
saying '0.004149' is not a valid floating point value.' I take this to 
mean that one of the points I made when digitizing is not "valid" and 
so tpsUtil cannot read the file? Is this right? When I was digitizing 
tpsDIG kept popping up error messages but I couldn't figure out what I 
was doing wrong, I was just tracing the outline using the 'draw curve' 
function. I'd delete the curve and start over, but it happened every 
time I outlined a molar so eventually I jut kept going. Did this mess 
me up? Was there a way to fix what I was doing that would have meant 
my curves would be read in tpsUtil? Is there a way I can manually 
append my curve data as landmarks that could still be read into 
Geomorph or Momocs as sliding landmarks? Could I edit the tps file 
manally to do this or am I going to have to start over redrawing the 
curves? I need to delete some landmarks as well, so I'm having the 
same problem with tpsUtil, I was planning on manually deleting the 
landmarks I no longer want for my specimens (or should I just tell the 
R packages I use to exclude certain landmarks? I think this might be 
better).


Any help would be greatly appreciated
Thanks
Candice Neves
MSc Candidate
--
MORPHMET may be accessed via its webpage at http://www.morphometrics.org
---
You received this message because you are subscribed to the Google 
Groups "MORPHMET" group.
To unsubscribe from this group and stop receiving emails from it, send 
an email to morphmet+unsubscr...@morphometrics.org 
.


--
Liu Idárraga O.
Estudiante Doctoral
Facultad de Ciencias Naturales y Museo
Universidad Nacional de La Plata
División de Mastozoología
Museo Argentino de Ciencias Naturales "Bernardino Rivadavia"
Av. Angel Gallardo 470
C1405DJR Ciudad Autónoma de Buenos Aires - Argentina
Phone: +5411- 4982 8370/6670 Ext. 209

--
MORPHMET may be accessed via its webpage at http://www.morphometrics.org
--- 
You received this message because you are subscribed to the Google Groups "MORPHMET" group.

To unsubscribe from this group and stop receiving emails from it, send an email 
to morphmet+unsubscr...@morphometrics.org.

Re: [MORPHMET] Save aligned in TPSRelW

[Liu Idarraga Orozco]

I use tpsRelw for align my data and I din't have ID lost in my matrix. 
Trait export the aligned file like a *tps. Respect to the size, I never 
has been "residuals of size" in there, and I think that you have a 
problem there, because the value should be "0" or nearly "0" (because 
the size was remove it). Cardini suggest: "


   As most of the time you may want to analyse both size and shape,
   instead of importing in MorphoJ the aligned specimens from the
   TPSRelw slid configurations, I'd suggest to save the SCALED ALIGNED
   SPECIMENS in TPSRelw."

but in my experience that funtion doesn`t work correctly with 
semilandmarks matrix, It creates a kind of loop for some landmarks. My 
suggestion is: save the aligned as *tps; and for the CS, I think a good 
option is import it to MorphoJ as a covariable, at least until the 
problem of "loops" is solved in tpsRelw.




On 22/06/17 07:16, andrea cardini wrote:
The point is simple: the variables called centroid size and log 
centroid size in MorphoJ won't be the real size of the original 
structures if they're imported without rescaling them.
They are, in my experience, about 1 and, regardless of the reason 
(which I find less important), they're wrong. Of course, if one does 
not do anything with size, shape data will be fine. I would not risk 
to be misled, nevertheless. In fact, I might have happened to review 
one or two papers when someone was testing allometry or something else 
involving size, when size was about 1 for all specimens.


Cheers

Andrea

On 22/06/17 11:18, Paolo Piras wrote:

Hi folks/Andrea,

just to underline this:

Andrea,when you say

"an 'apparent size variable' which is a series of numbers virtually 
equal to 1 (differences will be approximation errors)."


I think that MorphoJ "reads" the actual CS values from the aligned 
coordinates saved by TPSrelw that, as you say, have lost their 
original size. However, small deviations from 1 are not due to 
approximation error as I *suspect* that the "aligned coordinates" 
saved by TPSrelw are those projected on the tangent plane (by an 
orthogonal [more commonly used] or stereographic projection) and for 
this reason they have necesarily a CS larger than 1 that is the 
radius of the hyperspherical Kendall's Shape Space. In order to reach 
the tangent plane from the sphere one necessarily must go beyond its 
radius (that geometrically *IS* the size of a configuration lying on 
that space) and the CS will be inevitably larger than 1.
An indirect proof for this would be to verify that what you call 
"approximation error" gives CS values always slightly larger than 1 
and never smaller.
So I think there is no error in having, for those data, CS slightly 
different (=larger) from 1 as they MUST have CS larger than 1 by 
construction.

However,I'm not sure at 100% if TPSrelw works as I hypothesized here..
Playing a bit with TPSrelw could unveil this.
All the best
Paolo


2017-06-22 11:00 GMT+02:00 andrea cardini >:


Hi Pere,
there's plenty of work arounds this. I agree with Carmelo that the
nts format may be easier to edit manually, so that you can put back
the necessary labels.

HOWEVER, PLEASE BE CAREFUL because SIZE WILL BE MISSING from those
data. As most of the time you may want to analyse both size and
shape, instead of importing in MorphoJ the aligned specimens from
the TPSRelw slid configurations, I'd suggest to save the SCALED
ALIGNED SPECIMENS in TPSRelw. Jim added this option for this
purpose: this way you get the aligned specimens rescaled to the
original centroid size. Therefore, when you import them in MorphoJ
and redo the Procrustes fit, you will get both the aligned specimens
as well as their size.
In contrast, if you import the aligned data (not rescaled), size
won't be there but you may be easily tricked and think it is there,
because MorphoJ requires that you redo the Procrustes fit and then
you'll get an 'apparent size variable' which is a series of numbers
virtually equal to 1 (differences will be approximation errors).
Use the scaled aligned and you won't have this issue.

Cheers

Andrea

On 22/06/17 10:07, Pere Ibáñez wrote:

Hello All,

I am using TPSRelW to slide semilandmarks. Once I have the
consensus, I use Save aligned to export the aligned coordinates
(including slid semilandmarks) as a tps file, that I want to
import into MorphoJ.

The problem is that the file that is exported has the IDs of the
specimens empty, with no information. And MorphoJ doesn't import
it without that information, so I have to type it one by one on
the tps file. Does anyone know how to export that tps file from
TPSRelW without losing the ID?

Thanks!

-- MORPHMET may be accessed via its webpage at
http://www.morphometrics.org
---