-
Notifications
You must be signed in to change notification settings - Fork 19
Description
Hi there,
I am writing to report an issue I encountered when using Foldcomp to download AlphaFold databases, and to ask for guidance on the recommended workflow in HPC environments.
I am running Foldcomp on a Linux HPC cluster where compute nodes do not have outbound internet access. When executing:
import foldcomp
foldcomp.setup("afdb_swissprot_v4")
the download fails with the following error:
httpcore.ConnectError: All connection attempts failed
httpx.ConnectError: All connection attempts failed
From the traceback, it appears that foldcomp.setup() performs an online HTTP HEAD request to check file size, which is blocked in our environment.
I would greatly appreciate your advice on the following points:
Is there an official way to manually download Foldcomp databases (e.g. on a login node or external machine) and then use them locally?
What is the expected directory structure so that Foldcomp can recognize a pre-downloaded database without calling setup()?
Is there a way to disable or bypass the online connectivity check and directly point Foldcomp to a local database path?
Do you have any recommended practices for running Foldcomp on HPC systems without internet access?
Foldcomp is extremely valuable for large-scale protein structure analysis, and we would very much like to integrate it into our local pipelines.
Thank you very much for your time and for developing such a useful tool.
Best regards,
Shang Gao
Institute of Crop Sciences,
Chinese Academy of Agricultural Sciences