We’re developing a network-based modeling approach that can be applied to scenarios where each subject has a matrix as input (e.g., correlation values among ROIs, white-matter properties among gray-matter regions, etc.). Here is a preprint for the modeling mechanism:
A new program for matrix-based analysis (MBA) will be available in the next AFNI release.
The underlying mechanism for MBA is theoretically discussed in https://www.biorxiv.org/content/10.1101/459545v1
MBA is run with a simple script (as shown in the examples in the program help) with input data formulated in a table that codes the associated regions and variables. The response variable is usually correlation values (with or without Fisher-transformation) or white-matter properties (e.g., fractional anisotropy, mean diffusivity, radial diffusivity, axial diffusivity, etc.).
Details about running MBA can be found in the program help:
MBA | less
Any feedback and suggestions are highly appreciated!
First of all, thanks for releasing MBA. I have been waiting for this.
I updated my AFNI and ran the command MBA.R | less (MBA | less gives an error “command not found”) to see how to use this program and this is what I got:
R CMD BATCH --no-restore --save --no-readline --vanilla /home/kabbas/abin/MBA.R --args /home/kabbas/abin/MBA.R
I was hoping you could help me with this.
We just did a build earlier today, but have noticed that “MBA” wasn’t copied into the compiled binaries (whoops).
We will do another build this evening, so that should have everything.
Thanks for bringing this to our attention.