What is a BIDS App?
What is a BIDS App?¶
Which container do I use to start building my BIDS App?¶
The only minimum requirements of a BIDS App’s container is its ability to run your pipeline. So for example, if your app is mostly Python based it should be sufficient to start with any image that has Python and include your pipenv dependencies.
I noticed that every BIDS app follows some specific arguments such as
participant_label. Can I add more arguments?¶
Yes, the mandatory arguments that you have listed should be in every BIDS App, but you are free to add more that are specific to the task your app will perform.
I noticed that there are 2 kinds of analysis levels –>
group. What does this mean?¶
Generally, “participant” means individual level analysis (for instance: single subject) The group level analysis can be thought of as the second step, where the input becomes the output of the “participant” level analysis. For example, generating statistic maps of each subject’s brain could be considered “participant”, while generating the average of these maps across the dataset could be considered “group”.
What do we do if our application does not have any use for the group level analysis?**¶
If your pipeline has no need for group level analysis, it is fine if it is only valid for the analysis_level argument (see fmriprep)
Is it mandatory to first check the dataset validity using the BIDS-validator?¶
It is an extremely helpful feature to have validation of the dataset as part of your tool. However, it’s not considered mandatory. (for instance: I know of many apps that just simply fail with an error message if the dataset is not BIDS compliant)
You can release BIDS Apps using your own or your lab’s account. However, if you want to be added to the BIDS docker hub, please message Chris Gorgolewski to have a repo created for you.
If you base your code on https://github.com/BIDS-Apps/example deployment on docker hub will happen automatically via circleci