dtm_tools

Temporary tools to help with comparing decision tree modularized/unmodularized for tedana.

How to set up

Instructions assume git repositories will be stored in ~/repositories, and data will be stored in ~/tedana_testing. You will need to modify paths if your setup does not accomodate this.

Clone the tedana repository and add my fork as a remote (see here) via

# if this path does not exist, make it
mkdir ~/repositories
cd ~/repositories
git clone https://github.com/ME-ICA/tedana.git
cd tedana
git remote add jbteves https://github.com/jbteves/tedana.git
git fetch jbteves JT_DTM

Install it into your current environment (you may want a different one than the one you have activated!)

cd ~/repositories/tedana
pip install -e .

Get this tool by running

cd ~/repositories
git clone https://github.com/jbteves/dtm_tools.git

How to Use

First, enter the repository for tedana.

cd ~/repositories/tedana

Change to a branch you'd like to test (most likely main) via

git checkout main

and then run tedana on a test data set, somewhere other than the repository, for example

cd ~/tedana_testing

and consider putting your data in a directory called test_data, like this:

mkdir test_data
cp DATA_FILES test_data/

And then run main using a tedana call like this:

tedana \
    -d DATA_FILES \
    -e ECHO_TIMES \
    --out-dir main_tedana_results

Then, re-enter the repository and change to the modularized branch via

cd ~/repositories/tedana
git checkout jbteves/JT_DTM

and then run tedana on a test data set. Use the existing mixing matrix (main_tedana_results/desc-ICA_mixing.tsv) with the --mix option in order to guarantee the same ICA components and save time. Use an output directory named dtm_tedana_results. The call should look something like

tedana \
    -d DATA_FILES \
    -e ECHO_TIMES \
    --out-dir dtm_tedana_results \
    --mix main_tedana_results/desc-ICA_mixing.tsv

After this completes, you can use the dtm tool via

python ~/repositories/dtm_tools/dtm_tool.py main_tedana_results/desc-tedana_metrics.tsv dtm_tedana_results/desc-tedana_metrics.tsv

For more options see

python ~/repositories/dtm_tools/dtm_tools.py -h

To run the kundu tree instead of the minimal tree, use the following call:

tedana \
    -d DATA_FILES \
    -e ECHO_TIMES \
    --tree kundu \
    --out-dir dtm_tedana_results \
    --mix main_tedana_results/desc-ICA_mixing.tsv