Run the pipeline without container

[giuliapeteani]

Hi, thanks for your work.

Could you please provide how to run the pipeline in case of manual installation of the environment?

If I run:

/home/user/RFdiffusion2/rf_diffusion/benchmark/pipeline.py --benchmarks 10_62
--num_per_condition 10 --num_per_job 2 --out run1/run1
--args "diffuser.T=20|50 diffuser.aa_decode_steps=5|10"
"diffuser.T=100|200 diffuser.aa_decode_steps=20|40"

where 10_62 is a item of interest in a json file named benchmarks.json, written following your example, I got:

Default apptainer not found (you can build it from exec/rf_diffusion_aa.spec): /home/giulia/RFdiffusion2/rf_diffusion/exec/rf_diffusion_aa.sif
No apptainer found. Attempting to run /home/user/RFdiffusion2/rf_diffusion/benchmark/pipeline.py with /home/user/anaconda3/envs/rfd2_env_124/bin/python
usage: pipeline.py [--help] [--hydra-help] [--version] [--cfg {job,hydra,all}] [--resolve] [--package PACKAGE] [--run] [--multirun]
                   [--shell-completion] [--config-path CONFIG_PATH] [--config-name CONFIG_NAME] [--config-dir CONFIG_DIR]
                   [--experimental-rerun EXPERIMENTAL_RERUN] [--info [{all,config,defaults,defaults-tree,plugins,searchpath}]]
                   [overrides ...]
pipeline.py: error: unrecognized arguments: --benchmarks --num_per_condition 10 --num_per_job 2 --out run1/run1 --args diffuser.T=20|50 diffuser.aa_decode_steps=5|10 diffuser.T=100|200 diffuser.aa_decode_steps=20|40

[rclune]

Here is the example demo command from the README:

apptainer exec --nv rf_diffusion/exec/bakerlab_rf_diffusion_aa.sif rf_diffusion/benchmark/pipeline.py --config-name=open_source_demo sweep.benchmarks=active_site_unindexed_atomic_partial_ligand

To run this via python it would be:

python /path/to/rf_diffusion/benchmark/pipeline.py --config-name=open_source_demo sweep.benchmarks=active_site_unindexed_atomic_partial_ligand

To run pipeline.py with your own JSON file it would look like this:

python /path/to/rf_diffusion/benchmark/pipeline.py --config-name=benchmarks sweep.benchmarks=10_62 <your arguments>

Unless you have defined num_per_condition, num_per_job, out and args elsewhere, they will not be recognized by the python script.

[giuliapeteani]

Thank you very much for your help.

It worked out in the end.

I would just like to inform you that, in order to run the entire pipeline (including the LigandMPNN and Chai1 steps) using my manually installed environment, I had to install several additional packages beyond those specified in the YAML file provided in the installation guide.
Moreover, I had to run Chai1 separately, as I started encountering compatibility issues related to the PyTorch version (specifically, between the pytorch version required by Chai1 and that installed in rfd2_env_124).