Command Line Interface

CLI

Getting started

The CLI use a minimal data model to ensure consistency across the tools offered by timagetk.

The data model is simple and only consist in:

  • a folders and files structure

  • a single JSON file that regroups the necessary metadata and follows the folders structure.

Semantic

  • An experiment is a collection of observations.

  • Each observation should have a raw dataset that contains a (multichannel/multi-angle) image or a time-series.

  • An observation should thus relate to a single biological sample.

File hierarchy

This is an example of the folders and files structure you should expect:

Experiment_A/
├── Observation_1/
│   ├── raw/                <- put your raw data there
│   ├── <dataset_name>/     <- created automatically using the CLI tools
│   ├── ...
│   └── Observation_1.json  <- define the basic metadata to perform tasks
├── Observation_2/
│   ├── raw/
│   └── Observation_2.json
:
└── ...

JSON file

The minimal JSON file to create should describe the object, the time (if temporal observation) and raw data.

For example, with the p58 multi-angle images time-series:

{
    "object": {
        "series_name": "p58",
        "species": "Arabidopsis thaliana",
        "observation_sample": "Floral Meristem",
        "NCBI id": "3702",
        "genotype": "wt"
    },
    "time": {
        "points": [
            0,
            26,
            49
        ],
        "unit": "h"
    },
    "raw": {
        "0": [
            "090223-p58-flo-top.lsm",
            "090223-p58-flo-tilt1.lsm",
            "090223-p58-flo-tilt2.lsm"
        ],
        "1": [
            "090224-p58-flo-top.lsm",
            "090224-p58-flo-tilt1.lsm",
            "090224-p58-flo-tilt2.lsm"
        ],
        "2": [
            "090225-p58-flo-top.lsm",
            "090225-p58-flo-tilt1.lsm",
            "090225-p58-flo-tilt2.lsm"
        ]
    }
}

Note that all file paths are relative to the JSON file and thus do not need to be added.

Also, 'object' description is not mandatory except for 'series_name'. However, it is highly recommended to improve the metadata associated to the experiments.