Sean

It was two queries. Running on a Intel Xeon Gold 6230 CPU, and 400 Gb of RAM. It's an HPC system where the foldseek databases are stored on a network...

I think I found approximately the code that does this for the public webserver. I guess I can probably figure out how to adapt that to a command line program:...

Ran into this issue today. Would be great to expose `--no-check-certificate` somehow.

I don't like the changes to `main`, but the changes to ` pgen_msa` are fine. If you revert the changes to `main`, I'll merge it.

Just to note that when we tested this, it doesn't seem to speed up sampling. Increasing to a batch size of 2 just makes each iteration take twice as long,...

What metrics are you trying to calculate? What errors are you seeing? * Single sequence metrics should work. Just upload the fasta file of the generated sequences to the "target_seqs"...

I've seen that happen for people running on ARM-based Macs. Conda doesn't have ARM compiled versions of some binary dependencies. Luckily those Macs can execute x86 code, you just have...

That's a good question. In theory, if you use git to checkout the code at the commit for: https://github.com/seanrjohnson/protein_gibbs_sampler/tree/v0.1.0 It should work with the docker instructions. However, it seems that...

Mostly dealt with in: b7e7119e737a4b19b5d66f58243d306dde843854

I hit the same thing today. The template really improves the model, but then the affinity prediction fails.