Morphometrics Software:
IMP-Integrated Morphometrics Package
Last Updated 5/03/04
IMP is an integrated 2-D Landmark-based
geometric morphometrics software package, with it's own file format system which
loads easily into software such as SPSS and Excel. IMP can also read and write
files in the TPS file format used by F.James Rohlf's TPS series software. Several programs for 3D data are also
available, as are programs to use semi-landmarks along curves.
IMP currently runs under Windows only.
If you are considering using IMP, Please read
WhatisIMP.pdf Version as of 1/17/02.
To use IMP, you will need to download the
self-extracting compressed mathematics and graphics libraries from Mathworks
(makers of MATLAB), needed to support IMP. This set of libraries is large, just
over 7 Meg, but it supports all of IMP.
Download: mglinstaller.exe
Version as of 2/5/01
If you want to use IMP and want all of the
basic IMP files in one bundle, you may want to download the combined,
self-extracting Zip File IMPBasics, which contains CoordGen, Regress, PCAGen,
CVAGen, TwoGroup and TradMorph, the core programs of IMP.
Download IMPBasics.exe
Version 6a
As
of 8/17/02
There are three categories of IMP
software: General Release Software
for which there are reasonably complete users manuals available, Undocumented
Software which has little or no documentation currently available and Beta
software, which is the 3D software which has not been used in a serious
research project to date. The
successful use of Undocumented Software or Beta Software most likely relies on
your having had direct contact with one of my direct collaborators, or myself.
I will eventually get all the users manuals written, but you are welcome to
download the software at any time. The
Undocumented Software is on this site primarily as a convenience for my direct
collaborators and their students and collaborators.
Individual IMP
files for Download:
General Release IMP
Files for Download
|
Program |
File |
Manual |
Description |
Latest Update |
|
CoordGen6 |
|
Also see docs below |
Coordinate Generating Utility, also file translator to/from X1Y1, TPS
file formats |
3/24/03 |
|
CVAGen6 |
|
Also see docs below |
Canonical Variates Analysis Program Try using PCAGen6 before using CVAGen, the 2 programs use the same
input files and operational format. |
3/25/03 |
|
PCAGen6 |
|
Also see docs below |
Principal components analysis based on partial warp scores, outputs
partial warps scores and principle axis scores |
7/10/03 |
|
Regress6 |
|
also see docs below |
Thin Plate Spline and Regression Utility |
3/25/03 |
|
SuperPoser |
-not included in the superpac file |
Allows for the exploration of a variety of alternative superimposition
criteria-different distance minimization criteria-Not recommended as a
general purpose tool, but provided as an alternative.....written by Dave
Liebner, Canisius ’03. |
3/7/03 |
|
|
TwoGroup6h |
|
Comparison of statistically significant differences in shape between
two groups |
7/10/03 |
|
|
TmorphGen6b |
Generates a set of traditional morphometrics variables from a set of
geometric morphometric measurements |
4/5/03 |
Additional Documentation On Plotting Updates Additional manuals, commentary, etc.
GridTrimming.pdf- A file on how to use the grid trimming feature added to most programs which display deformation grids.
ReferenceRotation.pdf- A file on how to use the reference orientation controls added 11/5/02 added to most programs which display deformation
grids
Undocumented
Software, IMP Files for Download
(Some of these actually have limited
amounts of documentation, but the users manuals are not currently adequate for
a general release. If you are familiar
enough with IMP and patient enough, you should be able to get these to work
satisfactorily.) These items are posted
here primarily for use by my direct collaborators.
|
Program |
File |
Manual |
Description |
Latest Update |
|
BigFix6 |
------ |
Averages landmarks across a baseline-special purpose preprocessing of
data |
3/17/01 |
|
|
GrowProfile6 |
----- |
Growth Profiles |
3/17/01 |
|
|
DisparityBox6 |
------ |
Disparity Box, designed for calculation of Morpological Disparity |
7/10/03 |
|
|
LMEdit6 |
|
LMEdit6 is an editting tool for altering landmark configurations,
allowing for averaging of measurements across an axis of symetry and
eventually other tasks |
6/17/03 |
|
|
MakeFan 3rd Beta Version (Works reliably, limited manuals) |
None Yet! |
MakeFan, a tool for drawing alignment “fans” at equal angular spacing
on an image prior to digitization-can be used with TPSDig, or as a really
primitive stand-alone digitizer |
7/10/03 |
|
|
|
|
Extra DLL needed for MakeFan-expand this into the Bin\win32 directory where MakeFan is installed, if MakeFan doesn’t work
properly initially |
|
|
|
PLSAngle Version 6b |
Partial-included in compressed file |
Partial Least Squares tool for comparing patterns of covariance between
blocks, using two data of 2 blocks each |
7/10/03 |
|
|
PLSMaker (was listed as PLSBlock, same program) |
PLSMaker _manual.pdf included |
Partial Least Squares Tool for studying patterns of covariance using
shape data |
3/23/03 |
|
|
SemiLand6 4th Beta Version |
Partial manual included in compressed files |
A tool for processing semi-landmarks, prior to use in other programs |
12/06/02 |
|
|
ShuffleAllometry6 |
----- |
Computes range of vector correlations that can be generated by random
shuffling of a vector |
2/03/01 |
|
|
SpaceAngle |
Limited manual |
Compares the angle between the spaces defined by the PC axis of
different groups |
8/19/02 |
|
|
Splice6 |
No manual, simple program |
A tool to splice together ASCII files, either IMP or TPS |
3/18/04 |
|
|
Standard6a |
Standardization Utility-removes linear dependences of shape on some variable |
8/19/02 |
||
|
TBox |
No manual yet |
Computes a t-test score, given the means, samples sizes and standard
errors in the measurements. Note the
use of the standard error, rather than standard deviation. |
7/10/03 |
|
|
VecDisplay6 |
Displays thin plate spline vectors for comparison purposes |
3/25/03 |
||
|
VecCompare6 |
----- |
Computes a growth angle between two groups. Resampling statistics. |
3/17/01 |
|
|
VecLand6 |
----- |
Displays landmark by landmark differences between two groups USE OF VECLAND NOT RECOMMEND-USE VECDISPLAY INSTEAD |
3/17/01 |
|
|
VecLand6a- Substantially altered version, with documentation! |
----includes manual- |
Displays landmark by landmark Differences in regress models between two groups |
2/10/04 |
IMP 3D Beta Programs
Below are a set of the basic programs needed to work with 3D landmark based morphometric data. Simple3D does file conversion and pairwise tests of differences in shape between two groups (ala TwoGroup6). Regress3 and ThreeDPCA carry out regression analysis and PCA using partial warp scores for three dimensional data. The three dimensional display is rather limited at the moment, but all the mathematics is in place. Take a look at these programs if you are interested and let me know what additions/changes/improvements might prove helpful. Brief manuals and example files are included with each of these programs.
|
Program |
File |
Description |
Latest Update |
|
Regress3D |
Regression analysis of 3D landmark data using partial warp scores. |
8/21/02 |
|
|
Simple3D |
File conversion, Procrurste Superimposition, Goodall’s F-test for differences in means, calculation of Procrustes distances between means for 3D landmark data. |
8/21/02 |
|
|
ThreeDPCA |
PCA for 3D landmark data, based on partial warp scores |
8/21/02 |
Update
Information:
3/17/01- There are a number of updates as of 3/17/01. Most notable is the
PCA program PCAGen6, which does Principal Components Analysis of data based on thin
plate spline decompositions of shape. Groups are color coded for display, and
the shape deformation implied by a pair of PCA scores can be illustrated. It
may also be used purely to generate thin plate spline scores.
Most of the rest of the software got relatively minor graphics upgrades,
including the ability to write .EPS files (encapsulated post script) that may
be loaded into Adobe Illustrator and the like.
3/28/01- I loaded an updated manual for PCAGen6, and a mild update of
PCGen6a, which now includes Scree plots of the eigenvalues for each PC, and a
statistical test of which PC axes are significantly different from one another.
4/01/01- Loaded a new version of PCAGen6, called PCAGen6b. There was an
error in the statistical test of PC axes significant in PCAGen6a, it reported 1
too many significant axes, sorry folks, load the new version.
6/06/01-Loaded a new version of Regress6, called Regress6c. This how has updates in the statistics (it
calculates significance levels for you now) and it also allows you of plotting
the reference form, the deformed shape (or target form), or both. TwoGroup6 also got some updates,
specifically the calculation of alpha levels from the F-values. Also CoordGen6b now includes the ability to
generate a RFTRA (Resistant Fit Theta-Rho Fitting) superimpositioning, if
that’s what you need!
8/01/01-Added a user supplied reference option to PCAGen6.
10/9/01 Put CVAGen up for downloads.
1/17/02 Updated Manuals
5/28/02 Fixed a couple of problems in TwoGroup6c and Regress6. TwoGroup6c will now output a “dummy coded”
file to be used in Regress6 to depict the differences between the mean form of
two groups. The problem copying the
Procrustes distances vs CS plots that occurred in Regress6 has now been fixed.
7/27/02 Added Updated versions of PCAGen6 and VecCompare6-More updates
soon!
7/31/02 Added updates of PCAGen6F, Regress6F and CVAGen6D, so that these
programs now round the reference form (to 5 decimal places) in exactly the same
fashion, so that they generate the same values for PW scores when generating an
reference form internally, or when loading a user reference. Earlier differences due to how references
were rounded and superimpositions were handled should now be eliminated.
All these programs now have the grid plus vectors display option,
PCAGen6F and CVAGen6D can now plot the mean PCA/CVA scores of each group on the
PC/CVA axes plots, CVAGen6 can now assign “unknown” specimens to a known group
based on CVA scores and has an assignment test for the validity of assignments. Regress6F has a permutation test based on a
form of a Generalized Goodall’s F-test for the significance of the regression
model. All of them also have the “Grid
Trimming” feature as well, which will allow better user control of the size of
the grid (which I still have to write the manual for! If you want to try it, let me know, I’ll try to write a quick
explanation.)
8/15/02- Added updates of PCAGen6g,CVAGen6e and Regress6g which have a
more stable calculation of the GLS Procrustes Mean form. The earlier versions had a subroutine which
could be unstable under some conditions, producing a reference with a centroid
sized that deviated from 1 (by several percent). This should not pose a severe problem for analyses done using
earlier reference forms, so long as the linear approximation to the curved
shape space is valid. If your data
occupies enough of shape space to depart from linearity, you should recheck
your results with the newer versions.
Sorry about the problem, it was an error that appeared erratically.
I have checked the PW scores and reference form produced against those
from TPSRegress, and they are currently in agreement.
Also, I have added an update of DisparityBox6d that allows for automated
loading of a sequence of files, and also has a variety of nearest-neighbor
calculations available within it. The
manual for DisparityBox needs updating though.
Added MakeFan (an alignment tool) and PLSBlock (a partial least square
tool), and an updated version of IMPBasics.
8/19/02 Added updates of CVAGen, CoordGen, PCAGen, Regress, DisparityBox,
Standard 6, which fixed an error in how these programs rotated specimens to
align with the PC axis of variation of the reference. In some situations earlier versions of these programs could
produced affine distortions of the data, particularly when the reference was
rotated substantially during the procedure.
DOWNLOAD THE NEW VERSIONS OF THESE PROGRAMS AND RECHECK YOUR
RESULTS. The distortion was affine and
applied evenly to all specimens, which makes it hard to pick out visually, and
probably means that it won’t effect inferences, but do recheck your
results. Thanks to Mark Webster at UC
Riverside for locating this problem.
I have also posted a semi-landmark tool, SemiLand6, which will allow
processing of semilandmarks placed along curve, and also PLSAngle for comparing
PLS\SVD vectors between groups, or comparing SVD axes to PC axis.
8/21/02 Added the three dimensional programs in Beta form.
8/23/02 Updated Three Dimensional Programs-Added a permutation F-test and
rudimentary plot controls.
8/27/02 Added better image controls to PLSMaker6. Added the pdf on grid trimming.
9/12/02 Posted the 2nd Beta versions of SemiLand 6 and
MakeFan. There was a change in the
algorithm used in SemiLand6 to estimate the local slope at the endpoints of
open-ended curves, which seems to improve the accuracy of semi-landmark
positioning when sliding semi-landmarks along curves. Additional image controls were also added to SemiLand6. Some additional features were added to
MakeFan to improve it’s utility as a digitizer, allowing user control of
landmark color and number, and the ability to erase fans.
11/11/02 Posted updated versions of PCAGen, CVAGen,Regress, PLSMaker,
PLSAngle and VecDisplay. The plotting
routines were updated, fixing an error in the routine that in some cases
produced an affine error in the plots (but not in statistical inference). The new routine also allows for accurate
user control of the orientation of the reference during display.
12/06/02 Posted an updated version of SemiLand6 which aligns
semi-landmarks to minimize Procrustes distance between semi-landmarks on a
curve and semi-landmarks on a reference form.
The minimization based on bending energy is not currently operating in a
stable manner, and so it has been disabled until such time as I can produce a
stable, consistent version.
1/14/03 Added a “comb” alignment tool to MakeFan6, as an alternative to
fans.
2/1/03 Minor alterations to the “comb” function in MakeFan6, to draw
perpendicular lines at the end of the comb.
3/7/03 Added versions of PCAGEN, CVAGen and Regress with a default
reference orientation setting, making the reference rotation optional, which
sound increase the speed of use. I also
put up the SuperPoser tool written by my student Dave Liebner which allows use
of a variety of robust error functions to be used in superposition (as
alternative approaches to resistant fit methods), for those who want to work
with this type of mathematical approach.
3/10/03 Added versions of VecDisplay, PLSMaker and PLSAngle with the
default reference orientation setting.
Also added a version of CoordGen which orients the Procrustes reference
form with the specified axis as close to horizontal or vertical as possible,
given the need to align the GLS Procrustes Reference with the principal axis of
the specimen.
3/14/03 Added the ability to display deformation grids and vectors
depicting the differences between the means of two groups of specimens to the
TwoGroup progam.
3/24/03 I have worked out an approach to the calculation of the uniform
component that does not require that the reference form be in principal axis
alignment. This approach is now
implemented in PCAGEN and PLSMaker.
Also, CoordGen now offers Procrustes Superimposition with or without
principle axis (PA) alignment.
3/26/03. Added revised versions
of Regress, PLSAngle, VecDisplay and CVAGen with the updated display routines,
which do not require principle axis rotation.
PCAGen was also updated to fix an error that crept into the last update
that caused the software to fail to reload data properly. Also, I made minor adjustments in the size
of the white space in the copy to clipboard and copy to eps file functions that
should eliminate or reduce some copy problems.
6/18/03. I have added an
“exaggeration” control into TwoGroup, and also have put up the lmedit6 tool
which can “edit”, renumber and reflect landmark configurations, and can also
“back-reflect” landmarks on one side of a bilaterally symmetric organism to
form symmetric landmark configurations.
7/10/03 I noticed that PCAGen
didn’t number specimens correctly when the axes are reversed (changing the axis
polarity), so I fixed that and posted a new version of PCAGen. There is also a new MakeFan up since the
last compile had errors in the line weight control and in digitizing, so get a
new copy of that one. There are new
copies of DisparityBox, TwoGroup and PLSAngle up which have the standard error
estimates for all statistics displayed.
There is also a new tool up called Tbox which calculates t-test scores
given the means, standard errors and sample sizes of two measurements. This is a bit different from the standard
t-test in that it takes the standard error as an input, rather than a standard
deviation.
Other
Software
Below are one or two other items, which
are fundamentally univariate tools I need to make available. When time allows, I’m going to put these on
a different page on the website, but for now, here they are:
walkpac.exe -> This is the Walkbox6 tool for testing
time series against the null model of a random walks, and contains some of the tests discussed
in Sheets, H. D.
and Michell C.E., 2001. Why the Null Matters: statistical tests, random walks
and evolution. Genetica 112-113 pg 105-125.
There is a pdf manual included that should be reasonable complete.
agropac.exe-> This is the GrowChoice software tool used in Zelditch, M.L, B.L. Lundrigan, H.D. Sheets and T.Garland, Jr. 2003. Do precocial
mammals
develop at a faster rate? A comparison of rates of skull development in Sigmodon
fulviventer and Mus musculus
domesticus. Journal of Evolutionary Biology, Volume 16, pgs not known at present. The manual may be a bit sketchy for this
program at the moment, let me know how it goes.
DefCat-> This is the DefCat retrodeformation operations test frame software, not currently intended for general distribution, collaborators only, thanks!
FMC-> This is the FMC Finite Mixture Coding Program for phylogenetic coding, test frame software, not currently intended for general distribution, collaborators only, thanks!
REGPAC6P-> This is an update of Regress6k called Regress6K+, it is a Beta version that checks the correlation of the linear tangent space distances
With the Procrustes distance in curved space. Includes a Word file called “Correlation of Euclidean and Procrustes” that explains how it works.
CorrExplore6-> This is a tool for exploring matrix correlations. Collaborators only, thanks.