Dear Enigma toolbox users,
I have implemented spin_test() in Matlab and Python and have noticed some little differences in p-value... Has anyone noticed that before?
I have attached the data I have worked with if you want to try it:
- expression_brainorder.csv: gene expression across 68 Desikan Killiany regions obtained and preprocessed via the
abagen toolbox
- workflow_matlab2.m in Matlab2021b: reads in my custom gene data preprocessed with abagen, uses random cortical thickness data, performs spin_test in addition to "normal" spatial Spearman correlation, and adjusts p-values for multiple hypotheses testing
- workflow_python.ipynb, a Jupyter Notebook doing the same
I'd appreciate any feedback since I am also not as familiar with Matlab, thank you!! :)
Best, Melissa