gPPI - Question about Interaction Regressors

AFNI Message Board
1

Hello!

apologies in advance for double posting! We are working on a gPPI analysis and wanted to verify we have the right interaction regressors.

For obtaining the interaction regressors, this is what is first suggested:

'First create a 1D (one column) file, conditionA.1D, with 0’s (at those TR’s where condition A does not occur), 1’s (at those TR’s where condition A occurs). Unlike the traditional PPI, no contrasting is performed here (thus no -1’s)."

To create this file, would something like timing_tool.py be sufficient? And then, would this file serve as the input, “one1.1D”, as stated in this section of the sample script: “head -${cc} stimuli/Allruns_stim_${cond}_time.1D |tail -1 > tmp.1D
waver -dt ${sub_TR} -FILE ${sub_TR} one1.1D -tstim cat tmp.1D -numout ${n_tp} > ${cond}${cc}${sd}.1D”

Thank you very much for your help!

2

Hi Prerona,

Yes, timing_tool.py can convert from timing files to binary 1D files on an arbitrary (and possibly upsampled) TR grid. There is a gPPI example that is a touch different from Gang’s under AFNI_data6/FT_analysis/PPI.
README.txt is actually the main processing script, maybe because I am weird.

There, timing_tool.py -timing_to_1D (with -min_frac) is used for the conversion you mention. It might be good to add -per_run_file to get a different 1D file for each run, so the head and tail commands might not be needed.

  • rick