Foldseek clustering stuck for days with errors

[YFeriel]

Hello,

I am using a protein structure database with Foldseek for clustering, but I encountered an issue with the following command:
foldseek cluster /data/foldseek/concat_db /data/db_clusters /data/cluster_tmp_dir

I received this error:
structurerescorediagonal /data/foldseek/concat_db /data/foldseek/concat_db /data/cluster_tmp_dir/4804289747088079168/pref /data/cluster_tmp_dir/4804289747088079168/pref_rescore1 --exact-tmscore 0 --tmsc>[=
Can not write to data file p
Can not write to data file /data/cluster_tmp_dir/4804289747088079168/pref_rescore1.29
...
Error: Rescore with hamming distance step died

Initially, I suspected it could be due to disk space or memory issues, so I added the --remove-tmp-files 1 option. While this resolved potential disk space concerns, the runtime increased significantly. The clustering process has now been running for over seven days and remains stuck at the same step:

structurerescorediagonal /data/foldseek/concat_db /data/foldseek/concat_db /data/cluster_tmp_dir/4804289747088079168/pref /data/cluster_tmp_dir/4804289747088079168/pref_rescore1 --exact-tmscore 0 --tmsc>[=

Could this error be related to disk space or memory limitations, or might it indicate a different issue? Also, are there any optimizations or alternative approaches you would recommend to reduce the runtime and avoid prolonged processing times like this?

Thank you for your help!

[martin-steinegger]

Did you resolve this issue? It sounds like a resource issue.