Having trouble with 3dDeconvolve glt comparison with update version of afni

Site-related Questions and Comments
1

Hi,
We are trying to run this code through a pipeline and we cannot figure out why it isn't working in the updated versions of AFNI. It doesn't seem to like the -gltsym portion of the script and says it doesn't recognize the -gltsym portion. However, this ONLY happens in AFNI 23-25 and not on our AFNI 22 (AFNI_22.2.06 (Aug 18 2022) [64-bit] to be specific) What are we doing wrong?

We have been trying to figure this out for over a week with 3 people looking at things. I am happy to send the full script if it's not immediately obvious why this isn't working.

		3dDeconvolve -input pb01.${subNum}-${sesNum}.r*.scale+tlrc.HEAD \
			-mask ${maskApply} \
			-censor ${filePrefix}_${censorFile} \
			-ortvec motion_demean.1D mot_demean \
			-polort 4 -float \
			-num_stimts 4 \
			-stim_times_AM1 1 stimuli/${stimFiles[0]} 'dmBLOCK' -global_times \
			-stim_label 1 baseline \
			-stim_times_AM1 2 stimuli/${stimFiles[1]} 'dmBLOCK' -global_times \
			-stim_label 2 equal \
			-stim_times_AM1 3 stimuli/${stimFiles[2]} 'dmBLOCK' -global_times \
			-stim_label 3 overConsec \
			-stim_times_AM1 4 stimuli/${stimFiles[3]} 'dmBLOCK' -global_times \
			-stim_label 4 exclConsec \
			-num_glt 3 \
			-gltsym 'SYM: +exclConsec -equal' \
			-glt_label 1 exclConsec > equal \
			-gltsym 'SYM: +overConsec -equal' \
			-glt_label 2 overConsec > equal \
			-gltsym 'SYM: +exclConsec -baseline' \
			-glt_label 3 exclConsec > baseline \
			-tout -fout -x1D X_${statType}.xmat.1D -xjpeg X_${statType}.jpg \
			-x1D_uncensored X_${statType}.nocensor.xmat.1D \
			-errts CGT_ErrorTS_${statType}.${subNum}-${sesNum} \
			-fitts CGT_FitTS_${statType}.${subNum}-${sesNum} \
			-bucket stats_${statType}.${subNum}-${sesNum}

Our 1D files are in start time and duration format like this...

0.000:7.281"
2

Hi-

Could you please copy+paste the run output, including the specific error message? That will greatly facilitate this process.

thanks,
pt

3

You've got redirection symbols in the GLT labels, and they are not protected by quotes. I suggest avoiding the '>' symbol, as well as spaces. Consider something like:

-glt_label 1 exclConsec-equal

-rick

4

We tried to modify the labels but it still is output this error...

** ERROR: 3dDeconvolve dies: Unrecognized command line option: '+exclConsec'
   Here's hoping these excerpts from '/root/abin/3dDeconvolve -help' enlighten:
        'expression must be enclosed in one set of'
        'response models, there are n-2 parameters'
        'response amplitude, the 'Rmodel' must be one of the 'dm''
5

Hi-

Can you please post your updated 3dDeconvolve command?

thanks,
pt

6

Yes, please post the entire command, and make sure that "-glt_label 2" is similarly fixed.

Thanks,

-rick

7

Here is what we put in this time around. It seems to keep breaking as we go into the glt portion?

3dDeconvolve -input pb01.${subNum}-${sesNum}.r*.scale+tlrc.HEAD \
		-mask ${maskApply} \
		-censor ${filePrefix}_${censorFile} \
		-ortvec motion_demean.1D mot_demean \
		-polort 4 -float \
		-num_stimts 4 \
		-stim_times_AM1 1 stimuli/${stimFiles[0]} 'dmBLOCK' -global_times \
		-stim_label 1 baseline \
		-stim_times_AM1 2 stimuli/${stimFiles[1]} 'dmBLOCK' -global_times \
		-stim_label 2 equal \
		-stim_times_AM1 3 stimuli/${stimFiles[2]} 'dmBLOCK' -global_times \
		-stim_label 3 overConsec \
		-stim_times_AM1 4 stimuli/${stimFiles[3]} 'dmBLOCK' -global_times \
		-stim_label 4 exclConsec \
		-num_glt 3 \
		-gltsym 'SYM: +exclConsec -equal' \
		-glt_label 1 exclConsec-equal \
		-gltsym 'SYM: +overConsec -equal' \
		-glt_label 2 overConsec-equal \
		-gltsym 'SYM: +exclConsec -baseline' \
		-glt_label 3 exclConsec-baseline \
		-tout -fout -x1D X_${statType}.xmat.1D -xjpeg X_${statType}.jpg \
		-x1D_uncensored X_${statType}.nocensor.xmat.1D \
		-errts CGT_ErrorTS_${statType}.${subNum}-${sesNum} \
		-fitts CGT_FitTS_${statType}.${subNum}-${sesNum} \
		-bucket stats_${statType}.${subNum}-${sesNum}
8

That's interesting. I don't see any bad characters or spaces after the end-of-line \. I could essentially run that command on my computer (using different files, modified from Bootcamp data).

Could you please copy+paste the full terminal text output of the command? That might help shed some light.

And just to check, you are running this in a bash shell (which I am guessing because an array variable is indexed with [0])?

--pt

9

Yeah, we are running this in bash. I'll pull the full script sent to AFNI this afternoon and send. We can't figure out why it keeps failing as we've referenced all the AFNI related documentation.

10

It would be good to see the new error messages, too.
Like for Paul, your script works for me.

-rick