Miscellaneous


XYOM, MISCELLANEOUS


Allometry

Input (2 files): Size & Shape

Data are entered in two separate files:

  • A one column file containing size (CS, PER, ARE, etc.)
  • A multicolumn file containing shape (PC, or DF preferably)
    • tangent space variables (cfr. orthogonal projections, ORP)
    • normalized elliptic Fourier coefficients (NEF)
  • Please do not forget to indicate which column (PC, or DF, or NEF) you want to examine on the output graph.

Analyses:

The linear regression is computed (on each column of your second input file).

Output:

  • Report: linear correlation coefficients, determination coefficients
  • Graph: Plot of Y on X (size), with predicted values in red.

Illustration

 


Comparisons of variances according to the Garcia (2012) method  

Input (2 files): Shape & Shape

Data are characters (generally shape variables) from which the covariance matrix is to be compared.

You will be asked to enter two files with same number of columns (use for instance the same number of first PC of each file)

Analyses:

For details and interpretation, please see : Garcia, C. A simple procedure for the comparison of covariance matrices BMC Evolutionary Biology 2012, 12: 222

Output:

Report,

Example:

Garcia et al. 2012: comparison of covariance matrices: Report

S1, S2 and S3 statistics are computed on the eigenvalues expressed as proportions of the total variance, and divided by 8 (the maximum for S1) so that they could vary between zero and 1.

S1 (general differentiation) is decomposed into:
S2 (difference in orientation)
and
S3 (difference in shape)

S1 0.04153369286655641
S2 0.030532958527033756
S3 0.01100073433952266

S1 5 % c. i.: from 0.03558030317506227 to 0.08834383776763419
S2 5 % c. i.: from 0.02887768619190432 to 0.07711978872395149
S3 5 % c. i.: from 0.00310489936622955 to 0.015467977480155125


Escoufier: Covariation of shape of two parts of the same body (eg. head and wing)

Input (2 files): Shape & Shape

Two files, having the same number of rows (individuals)
Variables are shape variables describing for each individual two (successive) different parts of the body
In both files the order of individuals is exactly the same

Analyses:

  • PLS (Partial Least Squares): silently (wait for next XYOM version to see the report)
  • Escoufier’s coefficient, with confidence interval (1000 bootstraps)

Output:

Report,

Example:

Escoufier coefficient: Report
Coefficient of Escoufier 0.14687039006109312
5 % confidence interval: from 0.062087525956944535 to 0.2902682784015856

Graphical output is in preparation (PLS related graphs)


Escoufier: Covariation of shape with external parameter (eg. head and latitude)

Input (2 files): Shape & Other (please use “Shape & Shape”)

WARNING: standardisation step necessary for external parameter data

Analyses & Output: as above


Metric disparity of landmark-based shape

Input: raw (coordinates of) landmarks

  • A single file containing raw coordinates of anatomical landmarks
  • Subdivision (if there is no subdivision, please enter the total number of rows)

Analyses:

  • GPA of each group
  • The sum of diagonal (trace) of variance-covariance matrix of shape variables (tangent space variables, or ORP) of each group.
  • Bootstrap to compare groups according to subdivisions
    (if there is no subdivision, please enter the total number of rows)

Output:

Report,

Example:

— — —
Metric Disparity, or Variance of Shape: Report
— — —
Variance of shape, group 1: 0.0017094319564931848

Variance of shape, group 2: 0.0018580320534635488

Variance of shape, group 3: 0.001345167825046605

Between group 1 and group 2: not S at P = 0.05
Between group 1 and group 3: not S at P = 0.05
Between group 2 and group 3: not S at P = 0.05
(1000 bootstraps, cfr. Zelditch et al, 2004)


Metric disparity of outline-based shape

Inputraw (coordinates of) pseudolandmarks

  • raw coordinates of contours (pseudolandmarks)
  • subdivision

Analyses: 

  • EFA of each group
  • The sum of diagonal (trace) of variance-covariance matrix of shape variables (normalised elliptic Fourier coefficients, or NEF) of each group
  • Bootstrap to compare groups according to subdivisions
    (if there is no subdivision, please enter the total number of rows)

Output: as above.

 

Repeatability: 

Input (2 files): Shape & Shape

Two files containing the same individuals. In both files the order of individuals is exactly the same.

Variables are shape variables computed from two (successive) digitisation sessions, either orthogonal projections after Procrustes superimposition (ORP), or normalised elliptic Fourier coefficients (NEF).

WARNING: for anatomical landmarks (LM), size and shape must be computed on the total sample of raw LM (the first digitisation followed by the second one). The two input files are then extracted (or “splitted”) from the size output (if it is about size repeatability), or from the shape output (if it is about shape repeatability). Please consider the illustration below:

 

WARNING: for size or for shape obtained from pseudolandmarks (outlines), there is no special need to merge the raw coordinates to produce (size and) shape variables from the two digitisation sessions. The normalized elliptic Fourier coefficients (NEF, or shape variables) are computed individually, they do not depend on the total sample parameters.

 

 


XYOM, MISCELLANEOUS: FILES UTILITIES


File Conversion

  • XYOM reads the CLIC format. While the input file for XYOM must follow the CLIC format, the output file(s) may be issued in different formats: CLIC, TPS, CVS or JSON formats.
  • To submit a TPS format to XYOM, you should first convert it to the CLIC format. This conversion will extract the landmarks and only the landmarks, with a very first row of free comment of your own.
  • If you want to convert a TPS format to the CLIC_DB format, just ask the conversion to be performed from TPS to CLIC_DB. Then, you can extract the CLIC format from the CLIC_DB or from the TPS.
  • To submit pseudolandmarks to XYOM, you must follow a special arrangement for the three first columns (number of points, x and y coordinates of the centroid point), see CLIC OTL format.
  • It is possible to convert a table of pseudolandmarks to the CLIC OTL format.

Selecting columns from a given file

Please enter the first and last columns of the sub-matrix to be extracted

A new file is created

Selecting rows from a given file

Please enter the first and last rows of the sub-matrix to be extracted

A new file is created

Concatenating files by rows

Input: 2 files, same number of columns (or not, if an outlines file)

Output: A new file is created

Concatenating files by columns

Input: 2 files, same number of rows

Output: A new file is created