Pre-processed but non-statistically analyzed data

Dear AFNI experts,

Could you please let me know how can I find pre-processed but non-statistically analyzed data from output?

Below is the script I used.
-anat_has_skull yes
-blocks despike tshift align tlrc volreg mask scale regress
-copy_anat mprage/mprage.{subj}.{cond}+orig
-dsets /{subj}/{subj}.{cond}/{subj}.{cond}+orig \ -tcat_remove_first_trs 3 \ -out_dir results \ -script py_result_{subj}.{cond}.script \ -subj_id {subj}
-tlrc_base MNI152_T1_2009c+tlrc
-tlrc_suffix _MNI152
-volreg_align_to third
-volreg_interp -Fourier
-volreg_opts_vr -twopass -noclip -dfile volreg.{subj}.{cond}.1d
-volreg_warp_dxyz 3
-mask_apply epi
-regress_stim_times /Volumes/BRAINSMITH_DS/Titus/AES2016_FNT/{subj}/{subj}.{cond}/timing/{subj}.${cond}.*.stim.txt
-regress_stim_labels F NF
-regress_basis ‘BLOCK(4,1)’
-regress_censor_motion 0.3
-gltsym ‘SYM: +F -NF’
-glt_label 1 F_vs_NF

And, the attached is the output files from this script. I am not sure if I can find the pre-processed but non-statistically analyzed data among these outputs or should run another script. Please let me know.


Hi, Jun-

Several places to look:

  • creates a commented (!) processing script automatically, so you can look through that for the exact outputs of things you want.

  • as part of its processing, commands including running an AFNI program called to go through and look for certain outputs, and then make a dictionary of them. You can see a description of all important dsets/info that it currently looks for with: -show_uvar_dict

That will allow you to find what “keys” in the dictionary you would be interested in, and you can then look in the file created by that program for your specific processing, which will be called: out.ss_review_uvars.json.
You can then see what specific file was created for that “key”.

  • you can look at the files in time reversed order:


and the largest files near the end are probably the processed time series you want: probably either errts* or fitts* files are what you are looking for (depends on whether you are performing resting state or task data, respectively).


Hi Jun,

Just to add on to Paul’s comment, note that the pb0* datasets are the EPI going through the stages of preprocessing. So the last one, pb04scale is the input to the regression (as you can see in the 3dDeconvolve command in the proc script). It should match the later created all_runs dataset, since there seems to be just one run. If you want to see tissue contrast, it might be worth looking at the volreg dataset.

As Paul noted, it is worth looking through the proc script a bit.

  • rick