Standard workflows

[dominikbach]

Feature Description

Background

Currently, users have to manually define workflows. This is to some extent unavoidable because a lot of the meta-data required for pre-processing and modelling is not provided in a standard format (PsPM or otherwise).

This provides a barrier for new users as well as for automated data analysis engines.

We have started writing standard pipelines for our own (published) data: https://gitlab.com/bachlab/PsPM_pipeline

This could be generalised if meta-data is available.

Tasks

1. Design data formats (BIDS integration #423) such as to allow automated workflows.
2. Automated pre-processing:

• detect data types contained in file
• gather meta-information required for pre-processing
• pre-process all raw data in file

3. Automated modelling:

• detect pre-processed data types contained in file
• gather task structure
• implement pre-defined model for this task structure, run for all modalities contained

Required input arguments:

• which events to model
• which modalities to process

Output:

• parameter estimates
• warning/error log

Separate pipelines for:

• fear conditioning (SCR, pupil, scan path speed, hpr)
• reward conditioning (hpr only)
• event-related modelling (SCR only)