AFNI gurujis--

My goal is use the -tpattern @filename option within afni_proc to do slice-time correction using the slice timings provided in a BIDS .json file. I've managed to confound my efforts in two easy steps:

First, read slicenums lines from a BIDS format .json file to get the slice-timing for the EPI data. Dump it into the 1D file slicetimes.1D

# first set a variable indicating the number of slices
set slicenums="`3dinfo -nk rest.nii`"
# next, read slicenums of slice-timing info from the .json
set slicetimes = "`less rest.json | grep -A ${slicenums} SliceTiming | grep -Eo '[0-9.]{1,}'`"
echo ${slicetimes} > slicetimes.1D

Second, call up afni_proc and, among several other things, ask it to read the slice-timing from slicetimes.1D

afni_proc.py -subj_id ${subID} \
-dsets rest.nii \
-copy_anat T1.nii \
-blocks despike tshift align tlrc volreg mask regress \
-tshift_opts_ts -tpattern @slicetimes.1D \
-mask_segment_anat yes \
-mask_segment_erode yes \
-align_opts_aea -ginormous_move -deoblique on -cost lpc+ZZ \
-tcat_remove_first_trs 4 \
-tlrc_NL_warp \
-tlrc_base MNI152_2009_template.nii.gz \
-volreg_align_to MIN_OUTLIER \
-volreg_tlrc_warp \
-volreg_warp_dxyz 3.0 \
-regress_motion_per_run \
-regress_censor_motion 0.3 \
-regress_censor_outliers 0.1 \
-regress_apply_mot_types demean deriv \
-regress_compute_tsnr yes \
-regress_anaticor_fast \
-regress_bandpass 0.001 0.198 \
-execute

The following error results:
** ERROR: mri_read_ascii: couldn't open file slicetimes.1D
** FATAL ERROR: Can't read tpattern file slicetimes.1D

The file is, indeed, readable in other contexts, though. I can 1dplot it and if I use 3dTshift the old-fashioned way, it works fine:
3dTshift -prefix foo -tpattern @slicetimes.1D rest.nii

Something with how the afni_proc wrapper wants to read the file, perhaps? Sorry. Hope I'm not doing something ridiculous but I'm at the bottom of my debug stack...

Oh, and: Precompiled binary linux_ubuntu_16_64: Jun 18 2019 (Version AFNI_19.1.21 'Caligula')

Thanks!