This repository contains Python code for calculating fMRI's brain's functional connectivity states as identified by clustering model.

Running should be done via main_FC_states.py script for dynamical FC, FC dynamics, FC states (kmeans or hidden markov model), and states features.

To get specific results for the states characteristics, running should be done via main_states_features.py. This will generate lifetimes and probabilities of states and it estimate the p-value between two conditions.

To run a main script, open your command line, navigate to the directory and run it according to the arguments.

python main_FC_states.py --input users/name/data/task1 users/name/data/task2 --output users/name/data/tasks_output --areas 66 --pca --clusters 4

python main_states_features.py --input users/name/data/concatenated_clusters.npz --output users/name/data/tasks_output/states --n_clusters 4 --starts users/name/data/tasks_output/starts.json --separate --clusters users/name/data/tasks_output/clusters.npz

Data should be stored either in .csv or .mat file formats. For each subject there should be a separate file (time x brain areas).

Two parameters are in the code as they are usually the same, however, check for for number of components in dim. reduction and TR in mean lifetime of states (default 2 for both).

Also, if running for different data sets, check the number of brain areas. Ideally, the number of brain areas and the brain areas themselves should correspond.

Questions about the code can be addressed at: kcapouskova[at]hotmail.com