Contrast builder for group analysis broken...looking for advice

I鈥檓 experiencing a reported bug in 1.8.4 docker referred to here: 馃悰 Error in building group config file by `cpac group feat load_preset` 路 Issue #1719 路 FCP-INDI/C-PAC 路 GitHub This is fixed for upcoming 1.8.5, but this doesn鈥檛 help me now.

This bug is preventing me from running any group analyses because it crashes while making the design matrices. I can make the contrast files design matrices manually, but then how do I ensure that it will be handled by the cpac pipeline later?
Also, as an aside: There lots and lots of inconsistencies in the online documentation and the command line documentation for cpac.

i鈥檓 doing the group analyses manually. perhaps warning users of 1.8.4 that group analysis builder is temporarily broken would be helpful, and perhaps a few hints how to proceed.

This is what I will do:

1st I used the instructions on this page to create the design files: GLM/CreatingDesignMatricesByHand - FslWiki

2nd, I will follow the steps manually included in the cpac code: Order of commands:
- Merge all the Z-map 3D images into 4D image file. For details see fslmerge <>_::

    fslmerge -t sub01/sca/seed1/sca_Z_FWHM_merged.nii 
                sub02/sca/seed1/sca_Z_FWHM.nii.gz ....  
        -t : concatenate images in time
- Create mask specific for analysis. For details see `fslmaths <>`_::

    fslmaths merged.nii.gz 
            -abs -Tmin -bin mean_mask.nii.gz
         -Tmin  : min across time
         -abs   : absolute value
         -bin   : use (current image>0) to binarise

- FSL FLAMEO to perform higher level analysis.  For details see `flameo <>`_::
    flameo --copefile = merged.nii.gz --covsplitfile = anova_with_meanFD.grp --designfile = anova_with_meanFD.mat 
           --fcontrastsfile = anova_with_meanFD.fts --ld=stats --maskfile = mean_mask.nii.gz --runmode=ols 
           --tcontrastsfile = anova_with_meanFD.con
        --copefile        : cope regressor data file
        --designfile      : design matrix file
        --maskfile        : mask file
        --tcontrastsfile  : file containing an ASCII matrix specifying the t contrasts
        --fcontrastsfile  : file containing an ASCII matrix specifying the f contrasts
        --runmode         : Interference to perform (mixed effects - OLS)
- Run FSL Easy thresh 
  Easy thresh is a simple script for carrying out cluster-based thresholding and colour activation overlaying::
    easythresh <raw_zstat> <brain_mask> <z_thresh> <prob_thresh> <background_image> <output_root> [--mm]
  A seperate workflow called easythresh is called to run easythresh steps.
.. exec::
    from CPAC.group_analysis import create_fsl_flame_wf
    wf = create_fsl_flame_wf()

Hello Joshua_Lee!

Sorry for the delayed response. Currently the Group Analysis feature in C-PAC is deprecated (since the switch from 1.7 to 1.8). We are working on fixing the issues and updating the code to the newest version of the participant pipeline to allow for modular use of the Group Analysis features. I am currently working on updating the features for this specifically. If you would like you I can share the fixes that I have been working on, but I have to tell you to proceed with caution as we need to do internal testing before we can release the scripts. If you have more questions over this, I can provide help on how to proceed with the analysis.



Thanks. I eventually gathered that was the case; I ended up using the extracted time-series.

Suggestions you might think about: a) clearly indicating in the release notes which features are deprecated, and/or b) issuing a stable release. While new features are fun and awesome, it can frustrate researchers trying to set up shop with cpac.

Thanks again for the hard work of the entire cpac team.