[RESOLVED] Hcp MSMALL output

General
1

Hi everyone,

I have a question regarding hcp_msmall output in qunex and I don’t know if it is just me not running the qunex command properly or not having the correct input. I am running the hcp_msall on some multi run ICA fix resting state data (from a HCP disease dataset) and not getting the correct output.

After running the pipeline in my rfMRI_REST I only have a rfMRI_REST_Atlas_hp0_clean_vn.dtseries.nii and not the:
rfMRI_REST_Atlas_MSMAll_hp0_clean.dtseries.nii
rfMRI_REST_Atlas_hp0_clean.dtseries.nii
rfMRI_REST_Atlas_MSMAll_hp0_clean_vn.dscalar.nii

However, in my fMRI_CONCAT_ALL folder I do have the following files created by the hcp_msmall pipeline:
fMRI_CONCAT_ALL_Atlas_MSMAll.dtseries.nii
fMRI_CONCAT_ALL_Atlas_MSMAll_hp0_clean.dtseries.nii
fMRI_CONCAT_ALL_Atlas_MSMAll_hp0_clean_vn.dscalar.nii
fMRI_CONCAT_ALL_Atlas_MSMAll_hp0.dtseries.nii
fMRI_CONCAT_ALL_Atlas_MSMAll_hp0_vn.dscalar.nii
fMRI_CONCAT_ALL_Atlas_MSMAll_mean.dscalar.nii

The resting state fMRI input files are called 1, 2, 3, 4 not rfMRI_REST1 etc and the qunex command ran successfully for all subjects (same with the ICA fix command).

I am wondering where the rest of the output is? Could it be that I have not called the command correctly? Or is it because the resting state fMRI files are just numbers not rfMRI_REST1 etc so qunex doesn’t know what type of file they (i.e to qunex they could be task or resting state) so the fMRI_CONCAT_ALL_Atlas_MSMAll outputs are actually the outputs of the hcp_msmall command?

Here is my msmall command

qunex_container hcp_msmall
–bind=“/share/HCP/HCP-EP/EP_study:/share/HCP/HCP-EP/EP_study,/share/ndadata:/share/ndadata,${HOME}/qunex_tmp:/tmp”
–sessionsfolder=“${STUDY_FOLDER}/sessions”
–batchfile-=“${STUDY_FOLDER}/processing/batch.txt”
–container=“${QUNEXCONIMAGE}”
–sessions=“${SESSIONS}”
–overwrite=‘yes’
–bash_pre=“module load qunex-img/0.100.0;module load cuda-img/9.1;module load matlab-uon/r2024a;unset TMPDIR;export TMPDIR=$HOME/quenx_tmp;”
–scheduler=“SLURM,time=24:00:00,ntasks=1,cpus-per-task=1,mem-per-cpu=50000,partition=${QueueName},qos=img,jobname=hcp_msmall${ID}”

And I am using qunex 0.100

Thanks so much!
Daniel

2

Hi Daniel,

There is a setting that defines how your BOLDs will be named. That setting can be used from create_batch => setup_hcp onwards:

--hcp_filename (str, default 'automated'):
    How to name the BOLD files once mapped into the hcp input folder structure.
    The default('automated') will automatically name each file by their number
    (e.g. BOLD_1). The alternative ('userdefined') is to use the file names, which
    can be defined by the user prior to mapping (e.g. rfMRI_REST1_AP).

If you set this to userdefined you should have the naming that you expected above (rfMRI_REST1_AP instead of BOLD_1). You can check what naming you have by going into the sessions HCP folder (<study>/sessions/<session>/hcp/<session>). If there are folders 1, 2, etc. there you are using the automated option, if there rfMRI... folders then you are using userdefined.

I think the outputs look correct:

  • Inside the rfMRI_REST folder are the concatenated results from all REST BOLDs, so if you have 4 rest BOLDs in your sessions the images in here will be a concatenation in time of all 4 REST scans.
  • Inside the fMRI_CONCAT_ALL you have a concatenation of ALL bolds, if your study only has rest data, these are again a concatenation of rest BOLDs.
  • Inside each BOLDs subfolder (1, 2, 3), you have cleaned up and MSMAll results for individual BOLD scans.

Best, Jure