Drops non-standard atom names from protein chains. Args: drop_oxt: If True, also drop terminal oxygens (OXT). Returns: A new Str

Encodes a single haiku param as bytes, preserving non-numpy dtypes.

Computes euclidean distance between 'vec1' and 'vec2'. Args: vec1: Vec3Array to compute euclidean distance to vec2: Vec3Array to compute eu

Extracts MSA features. Example: The input raw MSA is: `[["AAAAAA"], ["Ai-CiDiiiEFa"]]` The output MSA will be: `[["AAAAAA"], ["A-CDEF"]]` The

Extracts species ID from MSA UniProtKB sequence identifiers. Args: msa_descriptions: The descriptions (the FASTA/A3M comment line) for each of

Featurise the folding input. Args: fold_input: The input to featurise. ccd: The chemical components dictionary. buckets: Bucket sizes t

Featurises the templates and returns a map of feature names to features. NB: If you don't do any prefiltering, this method might be slow to run

Returns the absolute path to an external resource. Note that this calls resources.GetResourceFilename under the hood and hence causes par file un

Fill in the optional fields (atom_element, res_name, chain_type). Extracts the optional fields (atom_element, res_name, chain_type) from a flat r

Fill missing MSA and template fields with default values.

Fill missing MSA fields with default values.

Fill missing MSA and template fields with default values.

Filter a structure's atoms by a function of their coordinates. Args: coord_predicate: A boolean function of coordinate vectors (shape (3,))

Filter to one representative atom per polymer residue, ligands unchanged. Args: representative_atom_by_chain_type: Chain type str to atom n

Filter resolved residues using a boolean mask.

Returns the class for `module` and `name` if allowed.

Returns the res_name of the closest standard protein/RNA/DNA residue. Optimized for the case where a single residue needs to be converted. If re

Convert template features to AlphaFold 3 format. Args: template_features: Template features for the protein. num_res: The length of the ami

Flatten Struct of Array instance.

Describes the relationship between keys in the database. Returns: A map from table names to pairs of `(column_name, foreign_table_name)`

Converts 1D array to a list of strings with the given number of decimals. This function is faster than converting via Python list comprehension, e.

Compute fraction of protein residues that are disordered. Args: struc: A structure to compute rASA metrics on. rasa_disorder_cutoff: The th

Constructs ProteinChain from the AlphaFoldServer JSON dict.

Constructs RnaChain from the AlphaFoldServer JSON dict.

Constructs DnaChain from the AlphaFoldServer JSON dict.

Construct Rigid3Array from homogeneous 4x4 array.

Returns a Structure constructed from atom array level data. All fields except name and, res_id are optional, all array fields consist of a value

Make Template from batch dictionary.

Construct atom layout object from dictionary.

Construct batch object from dictionary.

Create an Atoms table with minimal user inputs.

Create a Residues table with minimal user inputs.

Create a Chains table with minimal user inputs.

Constructs ProteinChain from the AlphaFold JSON dict.

Constructs RnaChain from the AlphaFold JSON dict.

Constructs DnaChain from the AlphaFold JSON dict.

Returns a new instance based on a given inference result.

Returns a new instance based on a given inference result.

Instantiates a AtomConfidence from a json string.

Returns a new instance from a given json string.

Returns a new instance from a given json string.

Construct a Structure from a mmCIF string. Args: mmcif_string: The string contents of an mmCIF file. name: Optional name for the structure.

Constructs an instance of ChemicalComponentsData from an Mmcif object.

Initializes the MSA from multiple A3M strings. Args: a3ms: A sequence of A3M strings representing individual MSAs produced by diffe

Constructs Rigid3Array by finding transform aligning points. This constructs the optimal Rigid Transform taking points_from to the arrangemen

Returns Structure created from from arrays with a residue dimension. All unset fields are filled with defaults (e.g. 1.0 for occupancy) or unset/

Returns a minimal structure for the input sequences and bonds. The returned structure will have at least one atom per residue. If the residue has

A full name of the hit.

Extracts the indices of the atoms that participate in bonds. Args: mmcif: The mmCIF object to process. model_id: The ID of the model that t

Returns mapping from entity ID to its type or polymer type if available. If the entity is in the _entity_poly table, returns its polymer chain type

Returns Dtype for given instance of dataclass.

Computes extra MSA features from the description.

Returns a mapping from internal chain ID to the author chain ID. Note that this is not a bijection. One author chain ID can map to multiple inter

Cross chain aggregate ipTM.

Returns length for given instance of dataclass.

Return ligand-ligand inter-residue bonds. Args: struct: Structure object to extract bonds from. only_glycan_ligands: Whether to only includ

Get the Haiku parameters from a model name.

Computes the MSA for a given query sequence. Args: target_sequence: The target amino-acid sequence. run_config: MSA run configuration.

Returns the type symbol (element) for all of the atoms. Args: mmcif: A parsed mmCIF file in the Mmcif format. ccd: The chemical component d

Return polymer-ligand & ligand-ligand inter-residue bonds. Args: struct: Structure object to extract bonds from. only_glycan_ligands: Wheth

Return polymer-ligand interchain bonds. Args: struct: Structure object to extract bonds from. only_glycan_ligands: Whether to only include

Returns the MSA profile and deletion_mean features.

Stochastic conformer search method using V3 ETK.

Returns the oldest revision date.

Gets the names of the atoms in a given CCD residue.

Returns the resolution of the structure. More than one resolution can be reported in an mmCIF. This function returns the first one (in the order

Returns Shape for given instance of dataclass.

Returns a list of file path or None if the input is not a shard spec. Args: shard_spec: the specifications of the shard, e.g. /path/to/file@20.

Returns the structure's stoichiometry using chain_res_name_sequence. Note that everything is considered (protein, RNA, DNA, ligands) except for

Returns a Structure where all fields are per-chain. e.g. chains = struc.group_by_chain['chain_id'] Returns: A new Structure with one a

Returns a Structure with one atom per residue. e.g. restypes = struc.group_by_residue['res_id'] Returns: A new Structure with one atom

Convenience function to create a plausible deprotonation field. Assumes a pH of 7 and always prefers HE2 over HD1 for HIS. Args: residues: a

Guess the polymer type (protein/rna/dna/other) based on the residues. The polymer type is guessed by first checking for any of the standard prote

Get parameters for linear layer. Parameters will be at least float32 or higher precision. Arguments: inputs: The input to the Linear layer.

Determine whether the structure has at least one clashing chain. A clashing chain is defined as having greater than 100 polymer atoms within 1.1A