This is a fork of netneurolab/markello_spatialnulls/parspin/. All rights belong to the original authors.

parspin

This directory is a tiny little Python package that we wrote to be used throughout our analyses. These codebits aren't necessarily very generalizable (though, for example, parspin.burt could be used in other projects!), but are re-used at various points throughout preprocessing, analysis, and results generation so they exist here rather than in any one single script.

You can check out the docstring of each module or simply read below to get an idea of what they do:

parspin.burt

This module contains our implementation of surrogate map generation as in Burt et al., 2018, Nat Neuro. The functions are annotated with relevant portions of the methods text from that manuscript that described the methodology.

parspin.partitions

This module contains helper code for generating the partition labels used in the "Testing partition specificity" section of the manuscript. Since we were using multiple parcellation atlases / resolutions (Cammoun 2012 and Schaefer 2018) and several partitions (Yeo intrinsic networks and von Economo cytoarchitectonic classes) we needed some easy functions for creating + grabbing the relevant annotation files—and this module contains that code!

parspin.plotting

This module contains code for plotting and saving brains. The make_surf_plot() was designed for quick-and-dirty interactive visualization whereas save_brainmap() was used to create all the brain figures seen in the manuscript.

parspin.simnulls

Contains some helper and utility functions for loading and working with simulated data.

parspin.spatial

This module contains all the code used to actually create simulated brain maps as described in the manuscript. The primary function (create_surface_grf()) should be relatively to port and use in other analyses, as desired. Note that some of this code is attributed to Joshua Burt (see the References in relevant function doc-strings).

parspin.surface

This module contains code for generating surface distance matrices. This was important because we needed fast, easy-to-use code that would let us (dis)-allow travel along the medial wall that was, critically, written in pure-Python. Much of this code has been migrated to BrainSMASH for broader use!

parspin.surrogates

Common functionality for generating surrogate brain maps for the Burt-2018 and Burt-2020 methods. Since these methods both depend on the brain map of interest (and therefore surrogates must be generated independently for each map), we needed to make the functions parallelizable to speed up computation.

parspin.utils

As the name suggests, this module contains simple utility functions that were moderately useful at various points of analysis.