ssh netid@login.storrs.hpc.uconn.edu
On your computer
touch testcopy
scp testcopy netid@login.storrs.hpc.uconn.edu:/home/NETID
or
scp testcopy netid@login.storrs.hpc.uconn.edu:~
ssh netid@login.storrs.hpc.uconn.edu
nano testcopy- Gorgolewski et al (2013)
- Test-retest dataset of various functional localizers
- Stored in
/scratch/psyc5171/dataset1 - New BIDS level: session
sub-01/ses-test
- Already in BIDS form
- Metadata is at the study level
- Your output and related files will go in
/scratch/psyc5171/NETID
- Today: run a single subject afni_proc on the cluster
- Next week: run the entire dataset
- Fork and clone hw7
- Figure out slice timing (non-multiband)
- Create event timing files
- Write
afni_proc.pycommand - Copy to cluster
- As a job:
- Run
afni_proc.py - Generate proc script
- Run proc script
- Run
- Look at
dataset1/task-fingerfootlips_bold.json - How are the slices being acquired?
- Slice times start at 0 and increase
- If we checked, the first slice would be inferior (=ascending)
- But not monotonically (=interleaved)
- Slice code is
alt+z - No timing file needed: use
-tpattern alt+z
- Block design motor localizer (
fingerfootlips) - 15 second blocks
- Timing for all subjects/sessions in
dataset1/task-fingerfootlips_events.tsv - Create a timing file (think back to HW1 or last class)
- Which convolution function should be used with
3dDeconvolve?
'BLOCK(15)'
- This can be done with a shell command similarly to HW1
- Use
awk 'BEGIN { ORS=" " }; {print $1}'as part of the command
- Let's modify
log2times.pyfrom the last class
finger-footfinger-lipsfoot-lips
afni_proc.pygenerates processing scripts- The
afni_proc.pycommand is generated byuber_subject.py - Edit
ap.shto work with the current data - Paths for input data should start with
/scratch/psyc5171/dataset1
- Customize
sbatch_ap.shandsbatch_proc.sh - Push your hw7 repository
- Login to HPC
cd /scratch/psyc5171/NETID- Clone your hw7 repository on the cluster
To generate afni_proc.py script
sbatch sbatch_ap.shTo run the proc script:
sbatch sbatch_proc.sh