ktch - A Python package for model-based morphometrics

ktch is a Python package for model-based morphometrics with scikit-learn compatible APIs.

Overview

ktch implements the following morphometric analysis methods:

• Landmark-based methods: Generalized Procrustes Analysis (GPA) with curve/surface semilandmark sliding, thin-plate spline interpolation
• Harmonic-based methods: Elliptic Fourier Analysis (EFA) for 2D/3D closed curves, spherical harmonic analysis (SPHARM) and disk harmonic analysis (DHA) for surfaces
• Theoretical morphological models of coiling: Raup's model and the growing tube model
• File I/O: Support for standard morphometric file formats (TPS, CHC, NEF, SPHARM-PDM)
• Datasets: Built-in example datasets for learning and testing
• Visualization: TPS deformation grids, PCA variance plots (with optional plot extra)

All analysis classes follow the scikit-learn API (fit, transform, fit_transform), so you can plug them into your existing pipelines.

Installation

Python >= 3.11 is required.

From PyPI

pip install ktch

From conda-forge

conda install -c conda-forge ktch

Optional dependencies

Via PyPI:

pip install ktch[plot]  # matplotlib, plotly, seaborn for visualization
pip install ktch[data]  # pooch for remote dataset downloads

Via conda-forge:

conda install -c conda-forge ktch-plot  # matplotlib, plotly, seaborn for visualization
conda install -c conda-forge ktch-data  # pooch for remote dataset downloads
conda install -c conda-forge ktch-all   # all optional dependencies

Development installation

git clone https://github.com/noshita/ktch.git
cd ktch
uv sync

Quick start

Elliptic Fourier Analysis on 2D outlines

from sklearn.decomposition import PCA

from ktch.datasets import load_outline_mosquito_wings
from ktch.harmonic import EllipticFourierAnalysis

# Load outline data (126 specimens, 100 points, 2D)
data = load_outline_mosquito_wings()
coords = data.coords

# Elliptic Fourier Analysis
efa = EllipticFourierAnalysis(n_harmonics=20)
coeffs = efa.fit_transform(coords)

# PCA on EFA coefficients
pca = PCA(n_components=5)
pc_scores = pca.fit_transform(coeffs)

Documentation

See doc.ktch.dev for full documentation:

• Tutorials: Step-by-step guides for GPA, EFA, spherical harmonics, and more
• How-to guides: Task-oriented recipes for data loading, visualization, and pipeline integration
• Explanation: Theoretical background on morphometric methods
• API reference: Complete API documentation

Contributing and support

Bug reports, feature requests, and usage questions are all welcome via GitHub Issues.

See CONTRIBUTING.md for development setup and conventions.

License

ktch is licensed under the Apache License, Version 2.0.