# Filestructure

If you use one of COSEDA's import tool, your dataset will be in the required structure automatically. If you use your own custom workflow to record the data, ensure you follow the same structure:

```
project_folder/
├── project.cosedawsp
├── dataset_1/
│   ├── dataset_1.h5
│   └── dataset_1.ini
│   └── processing_step_a_timestamp1/
│   └── processing_step_a_timestamp2/
│   └── processing_step_b_timestamp1/
├── dataset_2/
│   ├── dataset_2.h5
│   └── dataset_2.ini
└── ... (additional dataset folders)
```

Over the course of the processing, each dataset folder will be filled with additional subfolders for each processing step, where where the processing artefacts go. Sometimes you may want to run a processing step multiple times to tinker with the parameters. Each of this runs will generate its own timestamped subfolder. While the HDF5 file is updated with the latest processing output, the artefacts remain untouched. You can come back any time to compare results from previous runs.