Repository of scripts for processing data from the cue reactivity project

These scripts require the following software:

• AFNI
• Python 2.7
• Matlab
• matlab package, VISTASOFT

Start with the following raw data files:

• cue_matrix.csv -behavioral data from the task
• t1_raw.nii.gz –subjects t1-weighted anatomical scan
• cue1.nii.gz –functional task data

from a terminal command line, cd to the directory containing these scripts. Then type:

chmod 777 *sh

to be able to execute them.

(1) coregister subject's functional and anatomical data in native space and estimate the transform to bring subject's data into standard group space

from a terminal command line, type:

./preprocess_anat.sh

this script does the following using AFNI commands:

• skull strips t1 data using afni command "3dSkullStrip"
• aligns skull-stripped t1 data to t1 template in tlrc space using afni command @auto_tlrc, which allows for a 12 parameter affine transform
• pulls out the first volume of functional data to transform to group (tlrc) space; this is done either by:

• co-registering anatomy and functional data in native space, then applying the func > anat transform and anat > tlrc transform in the same step, (doFuncAnatCoreg=1) or
• simply applying the anat > tlrc transform to the functional data (doFuncAnatCoreg=0). This assumes that the anatomy and functional data are already in good alignment in native space, either because the subject didn't move or because they've been manually aligned prior to running this script.

QUALITY CHECK: in afni viewer, load subject's anatomy and functional volume in tlrc space (files "t1_tlrc_afni.nii.gz" and "vol1_cue_tlrc_afni.nii.gz"). These should be reasonably aligned. If they aren't, that means 1) the anatomical <-> functional alignment in native space messed up, 2) the subject's anatomical <-> tlrc template alignment messed up, or 3) both messed up.

(2) run pre-processing steps on functional data

from a terminal command line, type:

./preprocess_cue.sh

this script does the following using AFNI commands:

• removes first 6 volumes from functional scan (to allow t1 to reach steady state)
• slice time correction
• motion correction
• saves out a vector of which volumes have lots of motion (will use this to censor volumes in glm estimation step)
• spatially smooths data
• converts data to be in units of percent change (mean=100; so a value of 101 means 1% signal change above that voxel's mean signal)
• highpass filters the data to remove low frequency fluctuations
• transforms pre-processed functional data into standard group (tlrc) space using transforms estimated in "preprocess_anat.sh" script
• saves out white matter, csf, and nacc VOI time series as single vector text files (e.g., 'cue_csf_ts.1D')

(3) get stimulus onset times and make regressor time series

from matlab command line, type:

createRegs_script

this script loads behavioral data to get stimulus onset times and saves out regressors of interest. Saved out files each contain a single vector of length equal to the number of TRs in the task with mostly 0 entries, and than 1 to indicate when an event of interest occurs. These vectors are then convolved with an hrf using AFNI's waver command to model the hemodynamic response.

Check out the output files: from a terminal command line, cd to output "regs" directory, then type, e.g.,:

1dplot food_cue_cuec.1D

** Note that the output files from this script are used for estimating single-subject GLMs and for plotting VOI timecourses.

from matlab command line, type:

saveRoiTimeCourses_script

and then:

plotRoiTimeCourses_script

to save & plot ROI timecourses for events of interest

from a terminal command line, type:

python glm_cue.py

this script calls AFNI's 3dDeconvolve command to fit a GLM to a subject's fMRI data.

from a terminal command line, type:

python ttest_2sample_cue.py

to use AFNI's command 3dttest++ to perform t-ttests on subjects' brain maps

from matlab command line, type:

doQA_subj_script 
&
doQA_group_script

to save out some plots that display head motion. Use doQA_group_script to help determine what the threshold should be for excluding subjects based on bad motion.

in matlab, run:

analyzeBehavior_script
& 
analyzeBehavior_singlesubject_script

to check out behavioral data