kids_rsfMRI_motion

Requirements:

Python 2.7 matplotlib numpy pandas scikit-learn scipy

• DATA.tar.gz contains the resting state data used in this project.
• Phenotypic_V1_0b_preprocessed1.csv contains the necessary phenotype data.
• conda_env_py2.yml can be used to create a conda environment that approximates the one used for this project.

Split-half reliability

• abide_motion_wrapper.py performs an analysis for a single set of parameters.
• SgeAbideMotion.sh and loop_abide_motion_qsub.sh were used to parallelize the above script over many parameter combinations on a Sun Grid Engine computing cluster.
• plotting_data.ipynb plots the results.

Predictive reliability

• runSVCMotionTest.py for each iteration specified, creates a classifier for diagnosis on one half of a data set and uses that classifier to predict diagnosis in the other half, saving out the classification accuracy and model.
• create_fc_fisher_z_csv_file.py helper function to create data used in the above.
• SgeAbideMotionCV_UO.sh and loop_abide_motion_qsub_array_UO.sh (along with abide_motion_parameters.tsv) were used to run the above python script for different combinations of parameters on a Sun Grid Engine computing cluster.
• plot_svc_motion_comparison.md plots the data generated by the above python script and documents the code used to produce those plots (in plot_svc_motion_comparison.R).
• cv_output contains all classification accuracy results.
• cv_models contains all classifier data.

Data file splits

Some data files were split using split, and so have filenames ending in "a[abcd...]". To join these files you can do cat FILENAME.tar.gz* > FILENAME.tar.gz.