|
Program/version |
Purpose and features |
Links |
Sage
Symmetry and Asymmetry in Geometric Data, version 1.05
(last compiled 09/17/08) |
- Sage accepts 2D landmark data (in X1,Y1,...,Xk,Yk; X1,Y1,...,Xk,Yk,CS;
or basic TPS formats) and decomposes variance into symmetric and asymmetric
components. The latter is further decomposed in directional and fluctuating
asymmetry.
- Procrustes ANOVA and MANOVA tests are performed to assess the significance
of symmetry (=individual), directional asymmetry, and fluctuating asymmetry
of shape and size (when applicable), given samples with at least two
replicates per specimen.
- Both Object (bilaterally symmetric structures) and Matching (bilaterally
symmetric parts) types of symmetry are handled.
- Covariance matrix correlations are also computed between symmetric
and asymmetric components of variation.
- Sage allows saving symmetrized datasets (average of relabeled, reflected,
and superimposed configurations), residuals from symmetric component,
plus shape configurations representing each component of variation (symmetric,
asymmetric, and error), plus expected covariance matrices from each
of these effects.
- A preliminary 3-D version of Sage is available
upon request only. Although it does most computations (ANOVA, MANOVA,
permutation tests), it is not yet capable to perform Procrustes superimposition
(you need to superimpose your data yourself, using, for example, David
Sheet's IMP software). Current version of Sage3D cannot yet produce
any visualization either.
|
Manual (PDF)
Screenshot
References
Version history
|
Mace
Matrix correlations for landmark data, version 1.01
(last compiled 03/31/07)
|
- Mace accepts three forms of data: covariance matrices obtained from
landmark data, actual landmark data (same formats as Sage), paired non-landmark
variables, such as partial warp scores, and unpaired (regular) data,
such as linear distances.
- Significance of matrix correlations is computed from permuting rows
and columns of covariance matrices, maintaining the pairing of X,Y coordinates.
Inclusion of covariance diagonals in permutation tests is optional.
- Significance tests include computation of repeatability, and bootstrap
estimates of correlation statistics.
- Covariance matrices can be transformed into an "isotropic" version,
in which all elements except for the diagonal are zero and all elements
in diagonals are equal, and a version in which variances are left as
observed and all remaining elements are set to zero. The resulting matrices
can be saved and reintroduced in the analysis.
|
Manual (PDF)
Screenshot
References
Version history
|
Mace3D
Matrix correlations for 3-D landmark data, version 1.0
(last compiled 08/31/06) |
- Mace3D has the same functionality as Mace, except that does not process
unpaired (traditional) data.
|
Manual (PDF)
References
|
SemiThinner
Utility to convert curves in semi-landmarks, version 1.0
(last compiled 09/08/06)
|
- SemiThinner does only one thing: accepts (2-D) TPS files with CURVES
statements and reduce the number of points to a specified number in
all specimens in the TPS file.
- SemiThinner allows saving only semi-landmarks, landmarks only, or
both landmarks and semi-landmarks in a XY... format.
- You can contact me if you need this program
to accept formats other than TPS, or 3-D data. Send me an annotated
sample of your format to add it to the program.
|
Manual (PDF)
Screenshot
|
Coriandis
Correlation analysis based on distances , version 1.1 beta
(last compiled 05/08/08)
|
- Coriandis provides a set of graphical and analytical tools to study associations among multivariate datasets (e.g. shapes), using distances among measured individual or species.
- Coriandis accepts 2D landmark, as well as non-landmark data, including distance matrices (irrespective of how they are computed).
- Referenced in Márquez & Knowles (2007) [PDF][abstract].
|
Manual (PDF)
Screenshot
References
Version history
|
Mint
Modularity and Integration analysis tool for morphometric data, version 1.0 beta
(last compiled 09/07/08)
|
- Mint is a tool for testing a priori models of morphological integration and modularity on multivariate data.
- Mint accepts general multivariate data, as a matrix with n individuals and m variables, as well as 2-D landmark data (XY and TPS formats).
- Models tested by Mint assign each variable or landmark to a module, and tests are carried out on the covariance matrix derived from a full set of models.
- Models can be loaded individually, in batch, or created and edited within Mint. All loaded/edited/created models are then tested simultaneously for relative of goodness of fit.
- Fit statistics are derived using a parametric approach and a resampling (jackknife) method.
- Mint also includes a tool to carry out a variant of Partial Least Squares analysis on which a partition or putative module is regressed against the full set of traits.
- Referenced in Márquez (in press) [abstract].
|
Manual (PDF)
Screenshot
References
Version history
|