Code used in the analyses for Sreekumar et al. 2018
The Jupyter notebooks and python scripts shared here are written in Python 2.7. They cover analysis of data from single trial betas to the figures included in the publication.
The array of permutations used is saved in multiple parts in data/input/hame_perms_all.npz_*. This file can be reassembled with the following command:
cat data/input/hame_perms_all.npz_* | gunzip -c > data/input/ham_perms_all.npz
Similarly, results from the ROI level analysis of the medial temporal lobe are saved in multiple parts in data/output/glm_roi/vishu_res/hame_roi_res.csv.gz_*. This file can be reassembled with the followig command:
cat data/output/glm_roi/vishu_res/hame_roi_res.csv.gz_* | gunzip -c > data/output/glm_roi/vishu_res/ham_roi_res.csv
We are unable to share all of the raw data in this study due to privacy concerns related to lifelogging data. We have saved the dissimilarity matricies in pickles in data/input/rsa_dataset_gps_time_old_exclude_drop_viv_bin_scan_time_rem_ham.pickle and rsa_dataset_gps_time_old_exclude_viv_bin_scan_time_rem_ham.pickle. These files were generated by the notebook Generate_dissimilarity_matrices.ipynb.
RSA analysis was carried out by for all permutations with anal\ExpSamp_SL_qwarp_trans_func.py. Examples of using that script are in anal\write_swarm_file.ipynb. The permutations were processed with anal\Process_permutation_results_all_models.ipynb. Tables and plots based on the processed results were generated with on of the notebooks Generate_peak_plots_culster_lists_* named for each model.
Single trial betas for the reminicence task are required to calculate the neural dissimilarity matrices. These are available in neurovault: https://neurovault.org/collections/JBMYPSYK/. In order to use the code in this repository, the individual niftis will need to be recombined into a single 4d nifti per subject.