TrajPy

Trajectory analysis is a challenging task and fundamental for understanding the movement of living organisms in various scales.

We propose TrajPy as an easy pythonic solution to be applied in studies that demand trajectory analysis. With a friendly graphic user interface (GUI) it requires little knowledge of computing and physics to be used by nonspecialists.

TrajPy is composed of three main units of code:

• Basic usage:
  • The GUI: it is where you interact with trajpy and the only thing you need to know to start using it
• Advanced
  • trajpy.py: it's the heart of trajpy, it computes the Features for characterizing the trajectories
  • traj_generator.py: a trajectory generator that can be used to build a dataset for trajectory classification

Our dataset and Machine Learning (ML) model are available for use, as well the generator for building your own database.

Installation

We have the package hosted at PyPi, for installing use the command line:

pip3 install trajpy

If you want to test the development version, clone the repository at your local directory from your terminal:

git clone https://github.com/ocbe-uio/trajpy

Then run the setup.py for installing

python setup.py --install

Basic Usage Example

Using the Graphic User Interface (GUI)

Open a terminal and execute the line bellow

python3 -m trajpy.gui

1 - You can open one file at time clicking on Open file... or process several files in the same director with Open directory...

2 - Select the features to be computed by ticking the boxes

3 - Click on Compute

4 - Select the directory and file name where the results will be stored

The processing is ready when the following message appears in the text box located at the bottom of the GUI:

Results saved to /path/to/results/output.csv

File formats

Comma separated values (CSV)

Currently trajpy support CSV files organized in 4 columns: time t and 3 spatial coordinates x, y, z:

t	x	y	z
1.00	10.00	50.00	50.00
2.00	11.00	50.00	50.00
3.00	11.00	50.00	50.00
4.00	12.00	50.00	50.00
5.00	12.00	50.00	50.00
6.00	13.00	50.00	50.00

See the sample file provided in this repository as example.

LAMMPS YAML dump format

LAMMPS YAML files are defined with the following structure:

    ---
    time: 0.0
    natoms: 100
    keywords: [id, type, x, y, z, vx, vy, vz, fx, fy, fz]
    data:
    - [1, 1, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, -nan, -nan, -nan]
    - [2, 1, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, -nan, -nan, -nan]
    - [3, 1, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, -nan, -nan, -nan]
    ...

We provide support for parsing this type of data files with the function parse_lammps_dump_yaml().

Scripting

First we import the package

import trajpy.trajpy as tj

Then we load the data sample provided in this repository, we pass the arguments skip_header=1 to skip the first line of the file and delimiter=',' to specify the file format

filename = 'data/samples/sample.csv'
r = tj.Trajectory(filename,
                  skip_header=1,
                  delimiter=',')

Finally, for computing a set of features for trajectory analysis we can simple run the function r.compute_features()

    r.compute_features()

The features will be stored in the object r, for instance:

  >>> r.asymmetry
  >>> 0.5782095322093505
  >>> r.fractal_dimension
  >>> 1.04
  >>> r.efficiency
  >>> 0.29363293632936327
  >>> r.gyration_radius
  >>> array([[30.40512689,  5.82735002,  0.96782673],
  >>>     [ 5.82735002,  2.18625318,  0.27296851],
  >>>     [ 0.96782673,  0.27296851,  2.41663589]])

For more examples please consult the extended documentation: https://trajpy.readthedocs.io/

Requirements

• numpy >= 1.14.3
• scipy >= 1.7.1
• ttkthemes >= 2.4.0
• Pillow >= 8.1.0
• PyYAML >= 5.3.1

How to cite?

If using TrajPy for academic work, please cite our methodological paper and Software DOI:

@article{10.1093/bioadv/vbae026,
    author = {Moreira-Soares, Maurício and Mossmann, Eduardo and Travasso, Rui D M and Bordin, José Rafael},
    title = "{TrajPy: empowering feature engineering for trajectory analysis across domains}",
    journal = {Bioinformatics Advances},
    volume = {4},
    number = {1},
    pages = {vbae026},
    year = {2024},
    month = {02},
    issn = {2635-0041},
    doi = {10.1093/bioadv/vbae026},
    url = {https://doi.org/10.1093/bioadv/vbae026},
    eprint = {https://academic.oup.com/bioinformaticsadvances/article-pdf/4/1/vbae026/56926570/vbae026.pdf},
}

@software{mauricio_moreira_2020_3978699,
  author       = {Mauricio Moreira and Eduardo Mossmann},
  title        = {phydev/trajpy: TrajPy 1.3.1},
  month        = aug,
  year         = 2020,
  publisher    = {Zenodo},
  version      = {1.3.1},
  doi          = {10.5281/zenodo.3978699},
  url          = {https://doi.org/10.5281/zenodo.3978699}
}

Contribution

This is an open source project, and all contributions are welcome. Feel free to open an Issue, a Pull Request, or to e-mail us.

Publications using trajpy

Moreira-Soares M., Mossmann E., Travasso R. D. M, Bordin J. R., TrajPy: empowering feature engineering for trajectory analysis across domains, Bioinformatics Advances, Volume 4, Issue 1, 2024, vbae026, doi:10.1093/bioadv/vbae026

Eduardo Henrique Mossmann. A physics based feature engineering framework for trajectory analysis. MSc dissertation. Federal University of Pelotas 2022, Brazil.

Simões, RF, Pino, R, Moreira-Soares, M, et al. Quantitative Analysis of Neuronal Mitochondrial Movement Reveals Patterns Resulting from Neurotoxicity of Rotenone and 6-Hydroxydopamine. FASEB J. 2021; 35:e22024. doi:10.1096/fj.202100899R

Moreira-Soares, M., Pinto-Cunha, S., Bordin, J. R., Travasso, R. D. M. Adhesion modulates cell morphology and migration within dense fibrous networks. https://doi.org/10.1088/1361-648X/ab7c17

References

Arkin, H. and Janke, W. 2013. Gyration tensor based analysis of the shapes of polymer chains in an attractive spherical cage. J Chem Phys 138, 054904.

Wagner, T., Kroll, A., Haramagatti, C.R., Lipinski, H.G. and Wiemann, M. 2017. Classification and Segmentation of Nanoparticle Diffusion Trajectories in Cellular Micro Environments. PLoS One 12, e0170165.