AlphaFold and friends on the HPC

Previous Topic
Next Topic

Basic workflow example (ssh)

AlphaFold and friends on the HPC AlphaFold on the HPC Basic workflow example (ssh)

A full instruction set for running a new protein through AlphaFold is described in this section.

1. Log in using your credentials (details)

2a. If running for the first time: set up directories

After this, copy the content of the job script to the nano program on the command line (details on how to work with nano), or copy it from your personal computer using WinSCP. Note that pasting in PuTTY is done with a right-click by default.

2b. If directories already set up: navigate to the alphafold/ directory

3. Create a new FASTA file

You can paste the contents inside the command line using right-click. More details on working with nano can be found here.

4. Modify the job script to use your FASTA file

5. Swap to the GPU cluster

6. Submit your job, check progress, wait
The state (Q) will change to R when running, F when failed, and C when completed (or crashed).
The standard output and error files are created in alphafold/.

7. Gather outputs

The outputs (more details) can be found in the newly created directory in runs/. Copy the results you need (mainly the ranked_0 to 5 .pdb files to your personal computer, using for instance WinSCP).

8. Generate visualizations

As described in an earlier topic, you can generate visualizations for this run using an extra python script. Load in the correct modules (AlphaFold and matplotlib) and run visualize_alphafold_results.py as shown below.

Previous Topic
Back to Lesson
Next Topic