The virtual_histology script allows to perform imaging transcriptomics of a neuroimaging file.
The current version can be ran after having checked that you have all necessary dependencies (see below, we recommend using conda and setup an environment as described in spatialnulls), and downloading the script virt_hist.py by calling
pyhton virt_hist.py -i /path_to_scan.nii[.gz] -n 1
or
pyhton virt_hist.py -i /path_to_scan.nii[.gz] -v 60
where -n allows to control the number of components for the pls regression while -v indicates the percentage of variance to quantify.
The input parameters for the script are:
- -i (--input): path to a neuroimaging file in NIfTI format (.nii,.nii.gz) to analyse. Please note that it is assumed that this file has the same dimensions as the DK atlas used from the script (1 mm isotropic, matrix size 182x218x182). If this isn't true the script will raise an error. To resolve this simply reslice your imaging scan to the appropriate dimension.
- -n (--ncomp): number of components to use in the pls regression. Please note that in pls regression the first component is not necessarily the one explaining most variance.
- -v (--variance): minimum explained variance by the pls components. Please note that the amount of variance will be used to calculate the number of pls components (
integer number) that explains at least the desired variance. - -o (--output): path where to save the results. if none is given, the results are saved in the same path as the original scan.
NOTE: this initial release (v0.0) is the one tested for development and publication and is as coded, while we're currently working on a first release (v1.0) which will have a refarctored/cleaner code and a complete documentation.
Additionally we are working on an easy installation via pypi to allow direct calling from the terminal and in scripts (e.g., bash scripts).
python >= 3.7