SAMRI (Small Animal Magnetic Resonance Imaging) — pronounced "Sam-rye" — provides fMRI preprocessing, metadata parsing, and data analysis functions and workflows. SAMRI integrates functionalities from a number of other packages (listed under the dependencies section below) to create higher-level tools. The resulting interfaces aim to maximize reproducibility, simplify batch processing, and minimize the number of function calls required to generate figures and statistical summaries from the raw data.
The package is compatible with small rodent data acquired via Bruker ParaVision.
To execute the examples below, actual small animal imaging data is required.
If dependencies were managed via Portage (e.g. on Gentoo Linux) mouse brain atlases may already be present under /usr/share/mouse-brain-templates, and test data under /usr/share/samri_bindata and /usr/share/samri_bidsdata.
Otherwise you will need to manually fetch, unpack, and move such data to the correct directories, with the following commands:
wget http://resources.chymera.eu/distfiles/mouse-brain-templates-0.5.3.tar.xz
tar xf mouse-brain-templates-0.5.3.tar.xz
mv mouse_brain_atlases /usr/share/
wget http://resources.chymera.eu/distfiles/samri_bindata-0.4.tar.xz
tar xf samri_bindata-0.4.tar.xz
mv samri_bindata /usr/share/
wget http://resources.chymera.eu/distfiles/samri_bidsdata-0.3.tar.xz
tar xf samri_bidsdata-0.3.tar.xz
mv samri_bidsdata /usr/share/
All listed examples beyond this one necessitate a BIDS-compliant NIfTI directory tree, as produced via the following command:
SAMRI bru2bids -o . -f '{"acquisition":["EPI"]}' -s '{"acquisition":["TurboRARE"]}' samri_bindata
This executes minimal preprocessing and FSL's MELODIC, and is very useful for fast, qualitative diagnosis of functional measurement quality and/or stimulation efficacy.
SAMRI diagnose bids
SAMRI generic-prep -m '/usr/share/mouse-brain-templates/dsurqec_200micron_mask.nii' -f '{"acquisition":["EPIlowcov"]}' -s '{"acquisition":["TurboRARElowcov"]}' bids '/usr/share/mouse-brain-templates/dsurqec_200micron.nii'
Depending on your preferred package manager you may choose one of the following methods:
SAMRI is available via Portage (the package manager of Gentoo Linux, derivative distributions, and installable on any other Linux distribution, or BSD) via the Science Overlay. Upon enabling the overlay, the package can be emerged:
emerge samri
Alternatively, the live (i.e. latest) version of the package can be installed along with all of its dependencies without the need to enable to overlay:
git clone [email protected]:IBT-FMI/SAMRI.git
cd SAMRI/.gentoo
./install.sh
Python's setuptools allows you to install Python packages independently of your distribution (or operating system, even).
This approach cannot manage any of our numerous non-Python dependencies (by design) and at the moment will not even manage Python dependencies;
as such, given any other alternative, we do not recommend this approach:
git clone https://github.com/IBT-FMI/SAMRI.git
cd SAMRI
pip install --user .
If you are getting a Permission denied (publickey) error upon trying to clone, you can either:
- Add an SSH key to your GitHub account.
- Pull via the HTTPS link
git clone https://github.com/IBT-FMI/SAMRI.git.
Python's setuptools allows you to install Python packages independently of your distribution (or operating system, even);
it also allows you to install a "live" version of the package - dynamically linking back to the source code.
This permits you to test code (with real module functionality) as you develop it.
This method is sub-par for dependency management (see above notice), but - as a developer - you should be able to manually ensure that your package manager provides the needed packages.
git clone [email protected]:IBT-FMI/SAMRI.git
cd SAMRI
echo "export PATH=\$HOME/.local/bin/:\$PATH" >> ~/.bashrc
source ~/.bashrc
pip install --user -e .
If you are getting a Permission denied (publickey) error upon trying to clone, you can either:
- Add an SSH key to your GitHub account.
- Pull via the HTTPS link
git clone https://github.com/IBT-FMI/SAMRI.git.
Many SAMRI functions which take multiple paths as inputs, rely on what we call BIDS-Iterator Inputs.
These are pairs of one filename template string and a list of dictionaries (which are internally used by functions adhering to this input standard in order to format the aforementioned strings).
BIDS-Iterator Inputs can be produced via the samri.utilities.bids_substitution_iterator() function.
The most precise specification of the dependency graph (including conditionality) can be extracted from the SAMRI ebuild. For manual dependency management and overview you may use the following list:
- argh
- joblib
- matplotlib (>=
2.0.2) - NumPy (>=
1.13.3) - pandas
- seaborn
- statsmodels
- Blender
- FSL (>=
5.0.9) - Bru2Nii
- mouse-brain-templates — as generated by the mouse brain atlases generator
- nibabel
- nipy (>=
0.4.1) - Nipype (>=
1.0.0) - PyBIDS (<=
0.6.5) - scikit-image
- SciPy
- ANTs
- AFNI
- nilearn
Needed for package testing:
- pytest
- SAMRI example binary data: download link
- SAMRI example BIDS data: download link
Needed only in conjunction with LabbookDB metadata management:
As this is a top-level data analysis package, tests cannot be reliably performed without example data. While such data is packaged and correctly depended on by SAMRI, the long duration of running the tests cannot be easily mitigated. We thus recommend running the tests locally before submitting a pull request, and pasting the log. Tests can be run and a log recorded via:
cd your/location/for/SAMRI/
pytest -vv
Additionally, SAMRI offers strong separation of its modules, and often it may suffice to only run the tests in the module you have worked on, and on the high-level plotting module:
cd your/location/for/SAMRI/samri/plotting/
pytest -vv
This can allow you to skip the very time-consuming tests of the samri.pipelines module.