neurodamus-sonata-template

Template with minimal SONATA circuit with 1 neuron morphology and emodel to be tested with neurodamus docker image

Hippocampus use case:

1. clone this repo

2. go to to_docker/manipulate_nodes directory

3. to undertand the contents of nodes_template.h5 run:

h5ls -rd nodes_correct.h5

which will give the tree of hdf5 file with the values we want to obtain from nodes_to_change.h5. We want to modify model_template and morphology for node population 'popA/0'

/nodes/popA/0/model_template Dataset {1}
    Data:
         "hoc:CA1_pyr_cACpyr_mpg141017_a1_2_idC_2019032814340"
...
/nodes/popA/0/morphology Dataset {1}
    Data:
         "dend-050921AM2_axon-mpg141017_a1-2_idC"

4. Then we can specify which morphology and emodel to overwrite using the script. First lets copy the template.

cp nodes_to_change.h5 nodes_modified.h5

Now we will use update_nodes.py to modify nodes_modified.h5's contents

The first argument is to give node file to edit, then we give one by one the dataset field to change and the value we want it to be.

python update_nodes.py {nodes_file} {hdf5_path_to_dataset} {new_value} {hdf5_path_to_dataset2} {new_value2}

which translates into:

python update_nodes.py nodes_modified.h5 /nodes/popA/0/model_template hoc:CA1_pyr_cACpyr_mpg141017_a1_2_idC_2019032814340 /nodes/popA/0/morphology dend-050921AM2_axon-mpg141017_a1-2_idC

Use this nodes.h5 file in circuit directory. Better to copy it there.

cp nodes_modified.h5 ../circuit/nodes.h5

or you can make a symbolic link in order not to duplicate files

5. Now we follow the instructions in neurodamus to pull its docker image, compile our mod files and run the simulation

docker pull bluebrain/neurodamus

will give us the docker image

you can check if it is there with docker image ls

6. Then we mount our files into docker which we put before in to_docker/

docker run --rm -it -v full/path/to/to_docker:/mnt/mydata bluebrain/neurodamus

This will instantiate the docker and your terminal will change to dockers. You can always exit the image with exit command.

7. Compile your mod files into neurodamus. Here we use hippocampus mod files and also common mod files for all circuits which I put under to_docker/ for you.

cd /mnt/mydata/
cp $NEURODAMUS_MODS_DIR/* hippocampus/mod/
build_neurodamus.sh hippocampus/mod/

this will create a new folder x86_64/ under mnt/mydata. Then we run:

export HOC_LIBRARY_PATH=/mnt/mydata/hippocampus/mod:$HOC_LIBRARY_PATH

8. Once it compiles the mod files and we export its path to HOC_LIBRARY_PATH, we can run the simulation from mnt/mydata/ directory:

./x86_64/special -mpi -python $NEURODAMUS_PYTHON/init.py --configFile=simulations/run_no_dynamics/simulation_config.json

Then results will be under simulations/run_no_dynamics

Useful Links:

1. Neurodamus Docker: https://github.com/BlueBrain/neurodamus/tree/main/docker
2. BBP mod files : https://bbpgitlab.epfl.ch/hpc/sim/models

Note: Docker container itself doesnt have vim or bluepysnap so those should be installed separately.

apt-get update
apt-get install vim

python -mvenv venv
source venv/bin/activate
pip install bluepysnap

then you can use the snap soma report visualization under simulations/

---

The development of this software was supported by funding to the Blue Brain Project, a research center of the École polytechnique fédérale de Lausanne (EPFL), from the Swiss government's ETH Board of the Swiss Federal Institutes of Technology.

Copyright (c) 2024 Blue Brain Project/EPFL