transmorph 0.0.1

Optimal transport-based tools for data integration.

#+TITLE: Transmorph (anciently WOTi)

This python package aims to provide an easy interface for integrating datasets using optimal transport (OT)- and Gromov-Wasserstein (GW)-based methods. We plan to extend the package beyond data integration, with additional OT-related unsupervised and semi-supervised methods. Warning: This package is still in an early stage. Feel free to open an issue in case of unexpected behvior.

• Installation

#+begin_src shell git clone https://github.com/Risitop/WOTi pip install ./woti #+end_src

• Examples

See three example notebooks in =examples/= directory.

• Usage

This package offers four main integration techniques, two based on OT and two based on GW. Both OT and GW comes in two variants, balanced (similar to [[https://github.com/rsinghlab/SCOT][SCOT]] technique) and unbalanced, using a quadratic program in order to estimate data points weights. These weights are chosen so that the weighted Gaussian mixture distribution is close to be uniform over the dataset.

Assuming two numpy arrays X and Y representing source and target datasets, WOTi can be used in the following way. First, create a Woti object. The scale parameter adjusts kernel bandwidth, and needs some tuning according to cloud sparsity.

#+begin_src python import woti

X, Y = ... # datasets, np.ndarray integration = woti.Woti(method='ot') #+end_src

Then, simply apply the integration method to project X onto Y.

#+begin_src python X_integrated = integration.fit_transform(X, Y) #+end_src

• Reference

https://www.biorxiv.org/content/10.1101/2021.05.12.443561v1

• Documentation

Work in progress.