Between-group analysis of DWI parcelated connectome

Hi, I have processed my HCP-like DWI data and I now have DWI parcelated connectomes for each participant for both “Matrix 1” and “Matrix 3” versions, for example these are the files for a given participant:

10001_Conn1_waytotnorm_log_HCP-MMP1.0-SIsubregions_MSA_matrix1_log.pconn.nii
10001_Conn3_waytotnorm_log_HCP-MMP1.0-SIsubregions_MSA_matrix3_log.pconn.nii

I would like to do a between-group analysis using PALM from these data. I have 52 participants, divided in 3 groups (group 1 n = 17, group 2 n = 17, group 3 n = 18). I developed a GLM model to do pairwise comparisons for all group combinations, and anothe GLM model to do a F-test. I was wondering if you have any suggestions on how to best implement this in PALM. For example, can I follow the HCP course practical “P11_TaskfMRI_Analysis_PALM_BurgessWinkler.pdf” and adapt it to run between-group analysis using DWI parcelated connectomes as inputs?

Thank you.

Estephan

Hi Estephan,

I will consult our DWI/PALM experts and get back to you as soon as possible.

Jure

Hi Estephan,

Yes, you can adapt what’s in the in PALM practical. You can also reuse the GLM files you’ve already generated for use in PALM - you’ll just want to name them properly to make sure QuNex recognizes them.

Here’s an example on how to set up your design flag, from the QuNex wiki (run_palm — QuNex documentation):
–design=“name:sustained|t:taov|f:fmain”
Your file names should be named like the following (and located in the path specified by the --root command or in your current working directory):

  • sustained_d.csv … design matrix file
  • sustained_eb.csv … exchangebility blocks file
  • sustained_taov.csv … t-contrasts file
  • sustained_fmain.csv … f-contrasts file.

I assume you already have a design file, but if not you can make one following the specifications in GLM - FslWiki e.g. create a csv file with 3 columns (since you have 3 subject groups) and 52 rows (since you have 52 subjects).

For the --palm_args flag I usually use “n:1000|accel:tail|transposedata|demean|twotail” instead of the default for group comparisons, however I would go through each option in the QuNex wiki to make sure these values meet your needs. Transposedata will be required based on how you’ve set up your input files (if the command fails with an error like “number of rows in the input file doesn’t match” then you might need to add transposedata as an extra argument).

The FSL PALM user guide provides additional details on the input files (PALM/UserGuide - FslWiki) if you want more details on them.

Excellent, thank you for the detailed response. I’ll give it a try and let you know if any questions come up.