Question:Incomplete run on multiple fasta sequences

[farreve]

What is your question? I am planning to use ColabFold on a batch of sequences and am currently trialing the program on just 3 of my sequences (file name: test.fasta). When I run 'colabfold_batch test.fasta outputdir', within seconds the Query 1/3 is 100% complete but makes no progress on the other 2 sequences. Instead, I get ongoing, periodic log updates such as:

2023-08-03 14:15:45,538 alphafold2_ptm_model_1_seed_000 recycle=1 pLDDT=53.9 pTM=0.564 tol=4

which continue for 1+ hours before I eventually abort the program. I can successfully view/access the completed outputs from my first Query in outputdir, but can't find anything for the other two sequences (confirming they were never run on).

Is there anything I need to be doing differently? What is causing this incomplete run?

Computational environment

• macOS

To Reproduce Steps to reproduce the behavior:

1. Installed localcolabfold according to steps posted on GitHub
2. Created FASTA file (test.fasta) of 3 test sequences
3. Ran colabfold_batch test.fasta outputdir on test file

Expected behavior I expected for the run to fully complete for each of my 3 samples and upload outputs into the outputdir.

[farreve]

Thank you in advance for your support!

[farreve]

Update on this--After 4 "recycles", the program writes a variation of:

2023-08-04 02:34:56,306 alphafold2_ptm_model_3_seed_000 took 34313.1s (3 recycles)

What is the runtime I should expect? The program has been running for ~19hours and is still on Query 1/3

[YoshitakaMo]

That's expected behavior - AlphaFold2's GPU acceleration only supports NVIDIA GPUs on Linux or Windows, so calculations on macOS will be prolonged.

[farreve]

thank you! that makes sense