Martin Steinegger

Thank you for reporting this. This was a bug in the `structurerescorediagonal` code. It did not respect the tmscore threshold properly. I fixed it now and the following command results...

AlphaFold/UniProt50-minimal provides only representative structures and lacks C-alpha atom information, whereas the AlphaFold/UniProt50 database supports cluster searches, enabling retrieval of all AFDB (AlphaFold Database) members.

@szimmerman92 could you please try the newest version (commit f629bbe108595ebf0886dd6974e9959a4d962af3)? I implemented a different strategy that avoids reallocations in the prefilter.

It is the amino acid similarity. Currently, we do not support to print out both.

Why do you try to create a database from a database?

@linda5mith we did not implement `--alignment-type` in the clustering workflow. However, you can apply `--tmscore-threshold` to apply TMscore.

You can adjust the `--max-seqs` parameter to get more hits.

Our scores are not symmetrical since our bias corrections are query structure dependent.

We do not support running foldseek with amino acids at the moment. For that, foldseek would need to predict the structures, which we believe is currently outside the scope of...

Could you please post a link to a search result?