create_database_from_sequences

biotite.application_v2.mmseqs.create_database_from_sequences(app: MMseqsLikeApp, sequences: Mapping[str, Sequence | str], prostt5_weights: SequenceDatabase[Any] | None = None) → SequenceDatabase[Any]

Create a database from sequences.

The molecule type is taken from the type of the given sequences, so that short sequences are not misinterpreted by the detection of the application. Plain strings are interpreted as protein sequences.

Parameters:

appMMseqsApp or FoldseekApp

The application to use. Nucleotide sequences are only supported by MMseqsApp, as Foldseek compares structures.

sequencesmapping of str to Sequence or str

Sequence identifiers mapped to the respective sequence. All sequences must be of the same type. For Foldseek the corresponding 3Di sequences are predicted using ProstT5.

prostt5_weightsSequenceDatabase, optional

ProstT5 weights for Foldseek. If omitted, they are downloaded.

Returns:

databaseSequenceDatabase

The created database.

Notes

A search on a nucleotide database requires the search_type option (see SearchType), as the application cannot infer whether the sequences should be aligned directly or translated beforehand.

Gallery

Homology search and multiple sequence alignment

Homology search and multiple sequence alignment

Mutual information as measure for coevolution of residues

Mutual information as measure for coevolution of residues