Navigating TACC

TACC Basics

Information about TACC for bioinformatics is sparse on the web; the BioITeam Wiki (search for it) has some info. TACC has user guides, and a very useful consulting system for users. You can also email me or anyone else at the Bioinformatics Consulting Group.

TACC Website

portal.tacc.utexas.edu

• User Guides
• Consulting Ticket System. I’ve been impressed with their speed in answering questions (even very basic ones!), and their willingness to help me with odd problems.

Disk Sizes

• $HOME (cdh shortcut)
  • 5GB on Stampede, 5GB on Lonestar5
  • Backed-up
• $WORK (cdw shortcut)
  • 1 TB across all TACC systems
  • Not backed-up
• $SCRATCH (cds shortcut)
  • 8.5PB (8,500,000GB) on Stampede, 5.5PB on Lonestar5
  • Not backed-up, files older than 10 days deleted with no warning

Modules

• Search for a module: module spider <search_term>
• List current modules: module list
• Load a module: module load <module_name>
• Swap modules: module swap <old_module> <new_module>

Queues

• normal: best for most job submissions.
  • Stampede: 48 hour limit, 16 cores per node
  • Lonestar5: 48 hour limit, 24 cores per node
• largemem: when you need lots of memory, these have 1TB.
  • Stampede: 48 hour limit, 32 cores per node
  • Lonestar5: ? (Not clear from user guide)
• development: good for quick, small tests.
  • Stampede: 2 hour limit, 16 cores per node
  • Lonestar5: 2 hour limit, 24 cores per node

Launcher Scripts

I recommend using launcher_creator.py. Information is available here: https://wikis.utexas.edu/display/bioiteam/launcher_creator.py (or you can search for “BioITeam wiki” and look under “Software”)

To use launcher_creator.py, you should probably add a line to your .profile file (see last section).

Stampede and Lonestar5 both use SLURM as their job scheduling systems.

• Submit a job: sbatch job.slurm
• Check status: squeue -u <username> or qstat
• Delete a job:
  • Specific job: scancel job_id (get job_id from qstat)
  • All your jobs: scancel -u <username>

Pre-rolled BioITeam Scripts

Not working on Stampede yet, but soon!

• BWA-MEM: map_BWAmem
• BLAST+: split_blast
• Trinity: assemble_trinity

Add BioITeam To Your $PATH

One way to do it (but keep it on one line):

echo 'source /corral-repl/utexas/BioITeam/bin/profile_ngs_course.bash' >> ~/.profile_user

More Nodes Doesn’t Make It Faster

[top]