[FEAT] new pipeline building workflow

[yongninglei]

Proposed Workflow for Building New Pipelineworkflow in LaunchContainer (LC)

Recently, I have been following this workflow below to build new containers within LC. I am using prfpipeline as an example to illustrate the process.

1. LC structure for the new pipelines

• Each big steps will have their own subfolders: prepare, run, qc etc
• In the prepare subfolder, I coded the data cleaning steps in to different py scripts and they must follow the orders I was using a numbered structure (e.g., 01, 02, 03, etc.) for clarity and reproducibility.
• For correct orphan sessions, or the part requires manual edit, I write into separate script:
  • This script performs the required cleaning, but it is not generic, it is some session specific.
  • It serves as documentation of how such cases are handled.

2. Code Structure and Utilities

• For common functionalities such as:
  • printing/logging
  • generating subject lists
  • force overwrite options
• I follow the existing LC conventions.
• I build Reusable helper functions are implemented in lc/utils.py and imported where needed.

3. Development Workflow

• Code under active development is placed in the tests/ directory.
• Once the implementation is stable and validated, it is migrated into the main LC codebase.

4. Documentation

• After integration into LC, we finalize:
  • workflow documentation
  • any special notes that needs to be acknowledged, for example prfanalyze folder strucure is hardcoded in the orignal prfanalyze-vista matlab code by Benson, so we have to respect that in the LC workflow

---

Open Question

Does this workflow align with the current LC development practices?
Any suggestions or improvements would be very welcome.

[yongninglei]

This can be checked at the new PR tests subfolders