[python] Separation of Positive and Negative Values in a Gradient Distribution Across the Left and Right Hemispheres

[Yiiiike]

Hello,

What I Want to Achieve and the Problem I Encountered

I encountered some issues while using the Brainspace tool to calculate functional and structural gradients for the same dataset. Specifically, when mapping the functional gradient and structural gradient one onto the brain, there were no problems. However, the mapping result for structural gradient two shows a clear separation of positive and negative gradient values between the left and right hemispheres: the left hemisphere consists entirely of positive values, while the right hemisphere consists entirely of negative values.

Some of my attempts

After verifying that both the structural and functional connectivity matrices are symmetric and do not contain any rows of zeros, I proceeded to compute group-level templates and then calculated individual gradients using the configuration described above. My GradientMaps configuration is as follows:
GradientMaps(n_components=10, random_state=0, alignment='procrustes', approach='dm', kernel='spearman').

My development environment

My runtime environment is: Ubuntu 18.04.6, conda 24.11.0, Python 3.10, Brainspace 0.1.16.

Other Information

I consulted the manual but did not find any steps I may have overlooked. I suspect that the issue might stem from differences in the pipeline parameters used for processing the structural and functional matrices, as I applied the same processing pipeline for both. However, I could not find relevant explanations for this in the manual.

Do you have any insights or suggestions on how to address this issue? Could you guide me on possible directions to explore in solving this problem?

Thank you very much for your help, and I sincerely apologize for taking up your time!