Running the connectivitymatrix portion of QSIrecon on the HBN

1

Hi,

bear with me, I am normally working on structural stuff only, but am starting to dabble a bit in dwi as well.

We are looking at the data QSIprep’ed data of the HBN (accessible for registered users here: https://fcon_1000.projects.nitrc.org/indi/cmi_healthy_brain_network/downloads/downloads_MRI_R11.html). The paper An analysis-ready and quality controlled resource for pediatric brain white-matter research | Scientific Data describes the preprocessed data. Now we have two questions:

  1. The preprocessed data includes results from running AFQ. Is this equivalent to running the pre-defined AFQ workflow within qsirecon? Or are there differences?

  2. Can we use the already preprocessed data to then run qsirecon on top to get connectivity matrices.

Thanks,
Melanie

1 Like
2

Hi @melanieganz,

The differences would be mainly due to pyAFQ updates in QSIRecon since the release of this paper.

Yes.

Best,
Steven

3

Thank you so much Steven!

Do you by chance know of an example in the qsirecon docs where we can see how to run just the connectivity matrix quantification on top of already existing AFQ runs, so when they have been done outside of qsirecon? I looked at the documentation, but couldn’t really find anything like that outside of the workflows.

4

Hi @melanieganz,

You would need a sufficiently dense whole-brain tractogram to do that, which I do not think you would have from the HBN-POD2 outputs. Thus I recommend running the mrtrix_multishell_msmt_ACT-hsvs workflow in QSIRecon.

It uses mrtrix tools like dwi2response dhollander, dwi2fod msmt, mtnormalise, 5ttgen hsvs, etc. You can see some of the parameters in the recon spec itself: qsirecon/qsirecon/data/pipelines/mrtrix_multishell_msmt_ACT-hsvs.yaml at main · PennLINC/qsirecon · GitHub

Best,
Steven

5

Ah, thanks for that pointer!

1 Like
6

Hi!

I am the one running the analysis, and we needed to use mrtrix_multishell_msmt_ACT-fast as mrtrix_multishell_msmt_ACT-hsvs requires freesurfer files which are not available to us.

I still have an issue running this:

singularity run --cleanenv \
        --containall \
        --writable-tmpfs \
        -B /home/gkf879/license.txt:/opt/freesurfer/license.txt \
        -B ${INPUTS_DIR}:/inputs \
        -B ${RESULTS_DIR}:/results \
        -B ${WORK_DIR}:/work \
        ${SINGULARITY_IMAGE} \
        /inputs \
        /results \
        participant \
        -w /work \
        --nthreads 8 \
        --recon-spec mrtrix_multishell_msmt_ACT-fast \
        --participant-label ${PARTICIPANT}\
        --fs-license-file /opt/freesurfer/license.txt \
        --atlases AAL116 \
        -v

and it gives me this error:

Downloading https://templateflow.s3.amazonaws.com/tpl-MNI152NLin2009cAsym/tpl-MNI152NLin2009cAsym_res-01_T1w.nii.gz
^M  0%|          | 0.00/13.7M [00:00<?, ?B/s]^M  1%|          | 69.6k/13.7M [00:00<00:38, 357kB/s]^M  2%|▏         | 303k/13$Sentry is attempting to send 1 pending events
Waiting up to 2 seconds
Press Ctrl-C to quit

I have also downloaded the atlas and added it to my PATH.

Any ideas would be much appreciated

Kind regards
Annike Kristensen

7

Hi @Annike, and welcome to neurostars!

Personally, I’d recommend running freesurfer on the anatomical data if you can, the streamlines do look really nice with HSVS.

For that, it looks like your cluster doesn’t have internet connection. See workarounds here: Which version is "stable"? - #6 by Steven

Best,
Steven

8

Thanks Steven! I am jumping in here just to say that I agree with your take on FS, but unfortunately Annike doesn’t have the time for that. She’s just doing a 4 month MSc project and we were delayed by almost due months due to the university blocking data access. So this is not an analysis for a paper, but a type of analysis for her to play with and learn how to use qsiprep and qsirecon to get familiar with the tools and to do a small scale analysis.

Since I am mostly familiar with structural and fMRI besides PET, I had thought this would be a straightforward small-scale project given that the HBN is distributed and even partially pre-rpocessed. But I am also stumped with the documentation and with actually getting anything to run, which is a bit unusual. :wink: But it’s all about learnign new tools.

I don’t think the internet is an issue, since we can get stuff down with wget, but we’ll look at tour fix. Thank you for that pointer!

9

Hi @melanieganz and @Annike,

No worries, completely understandable!

Sorry I should have been more clear. Sometimes compute nodes lack internet access, while login / head nodes can access it fine. This is common in clusters to protect against several jobs overloading the system bandwith.

Best,
Steven

10

Ah, good tip! I will check that with our sys admin. I usually don’t use that cluster. Thanks!