shapes is a Python package for traditional and geometric morphometrics.
It can be used in a Python interpretator, or in a GUI online at PhyloNimbus.com, where you can also digitize landmarks.
Uses include:
- Procrustes superimposition.
- Missing landmark estimation.
shapes can be installed from the Python Package Index:
pip install shapes
Read files from Landmark editor:
import shapes
lms = shapes.read('/path/to/file')
This returns a Shape object for that specimen.
Shape objects can be combined into ShapeSets:
shapeset = shapes.ShapeSet(name="Theropod skulls")
shapeset.append_shape(lms)
Once you've added all relevant Shapes to a ShapeSet, the landmarks can be aligned using Procrustes Analysis:
aligned_shapeset = shapes.procrustes_align(shapeset)
Note that this Procrustes alignment will work even if some landmarks are missing (in which case alignment will be based solely on the non-missing landmarks). This could lead to systematic biases in your dataset depending on the distribution of missing data. Caveat emptor...
Once the data set has been aligned as in the previous step, missing data can be interpolated using one of two methods.
Using symmetry (for landmarks that are part of symmetrical pairs):
paired_landmarks = [(1,4),(2,5),(3,6)]
symmetrically_reconstructed = shapes.missing.estimate_from_symmetry(
aligned_shapeset,
paired_landmarks)
Or using the mean of the aligned homologous landmark in specimens where it is present:
mean_reconstructed = shapes.missing.estimate_from_mean(aligned_shapeset)
Landmarks from all Shapes in a ShapeSet can be exported to csv:
aligned_shapeset.export('output_file.csv')
shapes is available under the MIT license.
White, D.E. (2016) Chapter 3 of The Evolution and Phylogenetic Analysis of the Dinosaur Axial Skeleton. PhD Dissertation, George Washington University.