The uploaded zip-file entails the data needed for and obtained through the dissertation titled 'Topic modelling for the stratification of neurological patients' as part of the programme 'MSc. in Statistical Data Analysis' at Ghent University. The study aimed at exploring the applicability of hierarchical stochastic block models on resting-state functional magnetic resonance imaging (RS-fMRI) to cluster participants with known neurological disorders. The data is structured in different folders and aligns with the folder structure of the GitHub-repository that contains the analysis scripts (https://github.com/wvechelp/hsbm_on_fmri). The GitHub-repository already covers some example data, while additional data and results can be found in this zip-file. Additional comments on the analyses are also provided in the analysis scripts. The original raw data is obtained through the OpenFMRI project (http://openfmri.org/, with label ds000030) and as a Stanford Digital Repository (https://purl.stanford.edu/mg599hw5271) (Bilder et al., 2016). It is obtained from the NIH Roadmap Initiative, as a result of the Consortium for Neuropsychiatric Phenomics (CNP) study (Poldrack et al., 2016). Throughout the study, data was collected through interviews and rating scales, self-report measures, neurocognitive exams (using both paper-pencil and computerised tests), and a variety of neuroimaging data. More specific information on the selection procedure of the participants can be found in the description of the data (Bilder et al., 2016) and the associated article (Poldrack et al., 2016). Among the available neuroimaging data, the RS-fMRI data have been collected by asking participants to remain relaxed, while keeping their eyes open (with scans lasting 304 s and an image being collected every 2 seconds (Poldrack et al., 2016)). This raw data was pre-processed by Rasero et al. (2019) to correct for motion and temporal alignment. Smoothing (6-mm full width at half-maximum Gaussian kernel), intensity normalisation, and a band-pass filter (between 0.01 and 0.08 Hz) were applied prior to the removal of linear and quadratic trends. Motion time courses, average CSF signal, and the average white matter signal were regressed out prior to data transformation into voxels with a volume of 3 mm x 3 mm x 3 mm. Ultimately, the functional atlas of Shen et al. (2013) was used to average the voxel signals per anatomical region of interest (ROI), resulting in a parcellation of 278 ROIs (and associated time series consisting of 152 measurements). From these ROI-specific time series, Rasero et al. (2019) generated 278 x 278 matrices with Pearson coefficients. References: Bilder, R. M., Poldrack, R. A., Cannon, T., London, E., Freimer, N., Congdon, E., Karlsgodt, K., & Sabb, F. W. (2016). UCLA Consortium for Neuropsychiatric Phenomics LA5c Study Stanford Digital Repository. http://purl.stanford.edu/mg599hw5271 and https://openfmri.org/dataset/ds000030/ Poldrack, R. A., Congdon, E., Triplett, W., Gorgolewski, K. J., Karlsgodt, K. H., Mumford, J. A., Sabb, F. W., Freimer, N. B., London, E. D., Cannon, T. D., & Bilder, R. M. (2016). A phenome-wide examination of neural and cognitive function. Scientific Data, 3(1), 160110. https://doi.org/10.1038/sdata.2016.110 Rasero, J., Diez, I., Cortes, J. M., Marinazzo, D., & Stramaglia, S. A.-O. (2019). Connectome sorting by consensus clustering increases separability in group neuroimaging studies. Network Neuroscience, 3(2), 325-343. https://doi.org/https://doi.org/10.1162/netn_a_00074 Shen, X., Tokoglu, F., Papademetris, X., & Constable, R. T. (2013). Groupwise whole-brain parcellation from resting-state fMRI data for network node identification. NeuroImage, 82, 403-415. https://doi.org/https://doi.org/10.1016/j.neuroimage.2013.05.081