Support for providing inputs to AlphaFold2 as files and *not* using the container API endpoints?

AI & Data Science NVIDIA NIM Models
,
1

The AlphaFold2 documentation wants us to use the API endpoints exposed by the container. Our current workflow provides the inputs as files to the run_alphafold.py script. Is there a supported way to use the NVIDIA AlphaFold2 container without drastically changing our existing, functional workflow?

2

@TonyLovesDevOps, welcome to the NVIDIA Developer Forums!

The AlphaFold2 NIM is designed to take the protein sequence input as a string. If you have a file of sequences, you could read the file and loop over the list of strings, submitting an inference request for each.

You can find a description of the NIM API here:

You could submit the request to the hosted endpoint as described here:

Another option is hosting the NIM locally. You can find an example information for running the locally hosted nim here:

Hope this helps!