Thanks Doug - so the residual file would be [number of subjects x number of vertices] matrix? Does it have to be in special file format?

So you will generate the sig.mgh file? I would recommend just running mri_glmfit using a design close to whatever you did to generate the directory structure, then just replace the sig.mgh file with the file of your choosing. The one snag may be the fwhm.dat file. This is a measure of the spatial correlation in the residual. Since it works off the residual, it depends on the actual design. If your analysis produces a residual, you can run mris_fwhm to get the fwhm.dat file, and then just run mri_glmfit-sim.

doug

On 06/02/2016 02:56 PM, Chung, Yoonho wrote:

Dear experts,

This is a follow-up question from my previous post.

http://www.mail-archive.com/freesurfer@nmr.mgh.harvard.edu/msg47287.html

If I am running stats with my own stats code to generate p-values and other parameters, what other files

do I need to generate in order to run 'mri_glmfit-sim'? I do not have the glmdir file (since I am not running mri_glmfit) so I would like to know which files need to be supplied with which flag/option, in place of the '--glmdir', for mri_glmfit-sim to run the cluster-based correction for my vector of p-values. An example command would be extremely helpful. Thank you in advance!

Best,

Yoon