Using the ABFE_workflow without a scheduler

[NiklasTR]

Hi @RiesBen and team - I am currently trying to get a minimal example of the process with one ligand to run on a local GPU workstation. Is there a way to run the full workflow without slurm as a scheduler?

For example:

cli-abfe -p data/CyclophilinD_min/receptor.pdb -l data/CyclophilinD_min/one_ligand/ -o cyclo_out_small -njr 6 -njl 1 -nr 1

Thank you!

[RiesBen]

Hi @NiklasTR ,
Currently not form the top layer, you would need to use the lower python layers in order to prohibit cluster mode:
https://github.com/bigginlab/ABFE_workflow/blob/main/src/abfe/orchestration/generate_scheduler.py

However it is not advisable to use the script not on an HPC cluster, as calculations are very expensive.

[NiklasTR]

Hi @RiesBen - Thank you for getting back to me here - I will take a closer look at the scheduler config.

A bit of context: We're running our workloads on a Ray Cluster on top of Kubernetes right now - we wrote a pythonic handler for OpenFE (which works well!) and are now looking into an approach to running ABFE.

From going through the repo, it seems like most of the rules consume the scripts that are located in
https://github.com/bigginlab/ABFE_workflow/tree/main/src/abfe/scripts

If I wanted to get started by running a single molecule per worker (no sharding of the lambda windows for now), do I understand correctly, that the best approach is to "simply" run the scripts in the sequence outlined within the rules?