The Beowulf Cluster at U.M.-St. Louis

• Home
• Using the Cluster
• Programs
  • Bayesian Dating
  • ClustalW
  • Lamarc
  • Migrate
  • Modeltest
  • MrBayes
  • PAML
  • PAUP
  • PAUP Ratchet
  • POY
  • QTL Cartographer
• Troubleshooting
• The Kellogg Lab
• UMSL Beowulf Site

PAML on the cluster

---

PAML version 3.14 is on the cluster. Note that PAML has not been modified to run in parallel, so this script will run PAML on a single processor.

1. Create your data and tree files and make a control file. Make sure these files have UNIX line breaks (use TextWrangler).
2. 
   
3. Upload (sftp - use Fugu) your files to the cluster, ideally to an empty directory. Each PAML analysis should have its own directory.
4. 
   
5. To run your job, move into the directory with the control, data, and tree files that you want to analyze and type "runjobs.paml". You will be prompted to give the type of analysis that you want to run (codeml, baseml, evolver, pamp, or chi2) and for a name to give your job (so that you can recognize it on the cluster).
6. 
   
7. When the run starts and finishes, an email will be sent to your UMSL email account (i.e., yourgatewayid@umsl.edu). Depending on the size of your data set it may take some time to run.

Note that the program you run in PAML will look for a control file with the default PAML control file name. For example, running codeml will require that you upload a codeml.ctl file along with your tree file and your data file.

Note also that if you are running several PAML analyses at once, you will need to place the files needed for each analysis in their own folders.

Miscellaneous

When using the Branch Site Models particular branches need to be specified in the tree file to see if they have an increased dN/dS relative to other branches in the tree. To do this, generate the tree in PAUP and export in Phylip 3.X format. Open the file in BBEDIT/Text Wrangler and "estimate" where the branch of interest resides. Insert #1 in this place. Check the placement of the #1 by opening the tree file in TreeView. Click on the icon with 12 in the center. #1 will be visible at the node you specified in the treefile, and will demarcate the upper limit of the branch to be tested. If the #1 is in the incorrect position, alter the PHYLIP treefile accordingly and reopen in TreeView.

Although involved, this method seems to work more reliably than using the tree file generated in the M0 output file that is recommended in the manual.