Metadata-Version: 2.1
Name: brainsmash
Version: 0.9.0
Summary: Brain Surrogate Maps with Autocorrelated Spatial Heterogeneity.
Home-page: https://github.com/murraylab/brainsmash
Author: Joshua Burt
Author-email: joshua.burt@yale.edu
License: UNKNOWN
Description: BrainSMASH
        ==========
        
        BrainSMASH (Brain Surrogate Maps with Autocorrelated Spatial Heterogeneity) is a 
        Python-based computational platform for statistical testing of spatially
        autocorrelated brain maps. At the heart of BrainSMASH is the ability to 
        simulate surrogate brain maps with spatial autocorrelation that is matched
        to spatial autocorrelation in a target brain map. Additional utilities are provided
        for users using Connectome Workbench style surface-based neuroimaging files.
        
        Exhaustive documentation can be found [here](https://brainsmash.readthedocs.io).
        
        Dependencies
        ============
        Installing BrainSMASH requires:
        
        - Python 3+
        - [numpy](http://www.numpy.org)
        - [scipy](https://www.scipy.org/)
        - [pandas](https://pandas.pydata.org)
        - [nibabel](http://nipy.org/nibabel)
        - [matplotlib](https://matplotlib.org)
        - [scikit-learn](http://scikit-learn.org/stable/index.html)
        
        If you wish to use the additional utilities provided for Connectome Workbench users, you must have
        [Connectome Workbench](https://www.humanconnectome.org/software/get-connectome-workbench) installed with the ``wb_command`` executable locatable in your
        system PATH environment variable.
        
        Installation
        ============
        ---
        
        BrainSMASH is most easily installed using pip:
        
            pip install brainsmash
        
        You may also clone and install the source files manually:
        
            git clone https://github.com/murraylab/brainsmash.git
            cd brainsmash
            python setup.py install
        
        License
        -------
        The BrainSMASH source code is available under the GNU General Public License v3.0.
        
        Reference
        ---------
        Please cite the following paper if you use BrainSMASH in your research:
        
        Burt, J.B., Helmer, M., Shinn, M.W., Anticevic, A., Murray, J.D. Generative modeling of brain maps with spatial autocorrelation. Neuroimage, 220 (2020).
        
        Core development team
        ---------------------
        * Joshua B Burt, Murray Lab - Yale University
        * John D Murray, Murray Lab - Yale University
        
        Contributors
        ------------
        * Ross Markello - Montreal Neurological Institute
        
        Change Log
        ==========
        ---
        
        * 0.9.0 Added multi-dimensional Spearman rank correlation to `mapgen.stats` module.
        * 0.8.0 Parallelization of surrogate map generation is now supported thanks to new contributions from Ross Markello!
        * 0.7.0 Added `geo.volume` method to compute 3D Euclidean distance matrix from an arbitrary set of voxel coordinates, per several requests.
        * 0.6.1 Surrogates maps are now de-meaned prior to returning (as the mean carries no information).
        * 0.6.0 Added `unassigned_value` kwarg to `cortex` and `subcortex`.
        * 0.5.2 Introduced a bug during the last bug fix.
        * 0.5.1 Fixed bug which caused distances to be written to file one-dimensionally.
        * 0.5.0 Updated `geo.subcortex` to have parallel structure with `cortex`.
        * 0.4.0 Replaced `geo.cortex` function with Ross' new implementation, in a backwards-compatible fashion.
        * 0.3.0 Added ability to set seed/random state in Base and Sampled classes.
        * 0.2.0 Added Ross Markello's implementation of Dijkstra's algorithm for efficiently computing surface-based distances.
        * 0.1.1 Fixed bug in NaN handling.
        * 0.1.0 Added goodness-of-fit metrics to stats module.
        * 0.0.9 Fixed bug in Sampled.sampled.permute_map().
        * 0.0.8 Relaxed nibabel version dependency.
        * 0.0.7 Removed console print statements.
        * 0.0.6 Fixed masked dense array handling.
        * 0.0.1 Initial beta release.
        
Platform: UNKNOWN
Classifier: Programming Language :: Python :: 3.7
Classifier: License :: OSI Approved :: GNU General Public License v3 (GPLv3)
Requires-Python: >=3
Description-Content-Type: text/markdown
