Open tsalo opened 3 years ago
Will probably wait on this until #522 is handled and 0.0.9 is released, just so we have some separation between these two sets of impactful changes.
summary of how this could work:
refactor nimare representations of studysets to be in line with the NiMADs specification. Study classes will have multiple Analysis classes, which will in turn contain conditions, weights, Images, and Points. The Images and Points classes will in turn keep track of data that is often loaded into nimare,
for example, when a CBMA kernel is applied to Points, there will be a memory representation that activation map that is associated with that group of points.
Summary
It looks like NIMADS (example) has evolved and stabilized significantly, so it would be good to update how NiMARE stores and loads its JSON files. This won't affect most users, since most folks use Neurosynth- or Sleuth-format files, but it's still important to be up-to-date.
Additional details
This is distinct from #218, which would overhaul how data are represented. Although at some point we're going to offload all of this to Neurostore or PyNIMADS or something.
Next steps
nimare.dataset.Dataset.__init__
nimare.io.convert_neurosynth_to_dict
nimare.io.convert_sleuth_to_dict