Does foldseek still work when there are only structure and not aa sequence?

[ManlioWu]

Hello! Does foldseek use the amino acid sequence information when structure searching? I want to know whether it still work if I remove the aa sequence information. That is, there are only backbone structures.

[ManlioWu]

Hello! Does foldseek use the amino acid sequence information when structure searching? I want to know whether it still work if I remove the aa sequence information. That is, there are only backbone structures.

In other words, you mentioned that you combine amino acid and 3Di socres with weights 1.4 and 2.1, in the method section. However, I cannot find the corresponding code in foldseek-analysis. So I have the querstion if you use the amino acid sequence when creating subsititution maxtrix.

[YTang906]

I also want to do structure only alignment though I noticed the authors mentioned '0: 3Di Gotoh-Smith-Waterman (local, not recommended)'... If someone could tell me why this is not recommended I will really appreciate! (I am a novice in protein structure filed so I can't understand this....)

[martin-steinegger]

@YTang906 we now implemented https://github.com/steineggerlab/foldseek/commit/f0de872f3ab84bbd5c173424a6633f0384f3adbd the 3Di only mode better. You could give it a try now.