Using Kong 2019 parcellation with 91k grayordinates cifti output

[roeysc]

Dear CBIG lab members,

I'd like to use the Kong 2019 parcellation on single subjects that were preprocessed with fMRIprep, yielding cifti files in 91k_fs_LR space (as described here).
However, I see that the priors for cifti data are defined in 32k_fs_LR space.
Should I resample the cifti BOLD data for 32k space before running the analysis code? If so, how could this be done? It seems like I should use the Connectome Workbench cifti-resample function, but I could not find the 91k sphere to be used for downsampling.

On the other hand, this post on NeuroStars suggests that the 91k_fs_LR output of fMRIprep, in fact reflects surface data of 32k vertices. If that's the case, then I could simply use the Kong 2019 as is. Correct?

Many thanks for you help and for making this useful code open source!
Best,

Roey