gratin 0.1.11

Random walk analysis tool using graph neural networks

Gratin (Graphs on Trajectories for Inference)

Gratin is a tool to characterize trajectories of random walks, i.e. motion driven by random fluctuations. This type of motion is observed at various scales and in a wide diversity of systems. While this package was developed for the purpose of analysing experimental data coming from photo-activated localization microscopy (PALM) experiments, nothing prevents it from being used on random walk recordings coming from other experimental setups and other domains !

To extract summary statistics describing trajectories, Gratin mixes two ingredients :

• an original neural network architecture using graph neural networks (GNN)

• a simulation-based inference framework

Warning

Gratin requires the pytorch-geometric package, whose installation depends on you CUDA version. Note however that you do not need CUDA to run Gratin, it works on CPU, it’s only a bit slower. See here to install it on your machine.

References

• Hippolyte Verdier, Maxime Duval, François Laurent, Alhassan Cassé, Christian Vestergaard, et al.. Learning physical properties of anomalous random walks using graph neural networks. 2021. : https://arxiv.org/abs/2103.11738

• Hippolyte Verdier, François Laurent, Alhassan Cassé, Christian L. Vestergaard, Christian G. Specht, Jean-Baptiste Masson A maximum mean discrepancy approach reveals subtle changes in α-synuclein dynamics. 2022 : https://doi.org/10.1101/2022.04.11.487825

Note

This project has been set up using PyScaffold 4.1.3. For details and usage information on PyScaffold see https://pyscaffold.org/.