I’m Sourav Kulkarni from Bachlab and we are working on developing automated data processing pipelines for processing neuro data. When we are working on publicly available datasets we find that there are a few necessary pieces of information that either not included in the datasets, or are included in a non-standard way across different datasets coming from labs. A standardized format would be helpful and we’ve drafted a set of guidelines around what we think would be the ideal data format for building high performance automated data processing pipelines.
We would be happy to get some feedback and suggestions from the community to improve this further, perhaps as a BIDS Extension Proposal.
Hello @s.kulkarni, and thank you @tsalo for flagging me in the convo!
I think our effort is indeed overlapping (in part), and I’m very happy to see our solutions - despite some differences - are also converging in terms of important missing metadata.
We are trying to code not only raw data, but also derivatives. At the moment, our proposal for raw data is moving towards not proposing more suffixes, but working at the level of columns and metadata, with more recommended fields to improve replicability. We’re currently takling the pandora’s box of derivatives, that we wanted to at least have a basic proposal about before making our BEP public and get feedbacks on.
I am currently away, but the team I’m part of meets every two weeks, and I already flagged this conversation to them to invite you to the next meeting. If you are at OHBM, I’d be happy to have a quick chat here, but otherwise please contact us at physiopy.community@gmail.com to schedule a meeting!