Dumps a json string with JSON compatible NaN representation.

Filters hits based on the filter config.

Find chiral centres and detect their chirality. Only elements listed in _CHIRAL_ELEMENTS are considered as centres. Args: mol: The molecule

Flattens an author naming scheme table consistently with res_ids.

Returns cheaply the first model ID from the mmCIF.

Get initializer for weights.

Searches for templates for a single protein chain.

Computes internal residue ID for ligand residues that don't have it.

Query the database shard using Nhmmer.

Query the database shard using Jackhmmer.

Sample a random seed for AlphaFoldServer job.

Raises an error if any of the keys are not column names.

Aligns sequences in A3M to the profile and returns the alignment in A3M. Args: a3m_str: A list of sequence strings. profile: A hmm fi

Applies the rotation followed by the translation to `coords`.

Builds a HHM for the aligned sequences given as an A3M string. Args: a3m: A string with the aligned sequences in the A3M format. Retur

Compute predicted errors for all pairs of chains. Args: num_tokens: The number of tokens (not including padding). asym_ids: The asym_ids (a

Concatenates structures along the atom dimension. NB: By default this function will first assign unique chain IDs to all chains in `strucs` so th

Returns a shallow copy with the atoms table updated.

Provides a zero-indexed enumeration of residues in an atom iterable. Args: atom_iter: An iterable of atom dicts as returned by Structure.iter_a

Extracts embeddings from model outputs.

Extracts inference results from model outputs.

Featurises the MSA and returns a map of feature names to features. Returns: A dictionary mapping feature names to values. Raises:

Replaces non-standard polymer residues with standard alternatives or UNK. e.g. maps 'ACE' -> 'UNK', 'MSE' -> 'MET'. NB: Only fixes the resid

Creates an AtomLayout object from a numpy array with shape (6, ...). see also to_array() Args: arr: np.ndarray of object with shape (6,

Constructs an instance of `BioassemblyData` from an `Mmcif` object.

Loads the input from an mmCIF string. WARNING: Since rng seeds are not stored in mmCIFs, an rng seed is sampled in the returned `Input`.

Construct Rot3Array from components of quaternion.

Returns hits + Structures, Structures filtered to the hit's chain.

Returns an mmCIF for a given `target_name`. Args: target_name: Name specifying the target mmCIF. Raises: NotFoundError: If the t

Gets the row at the specified index.

Gets the table with the given name.

Returns a hash ignoring the ID - useful for deduplication.

Iterates over the atoms and bond information. Example usage: ``` for from_atom, dest_atom, bond_info in struc.iter_bonds(): print(

Iterates pairs of (chain_start, chain_end) indices. Yields: Pairs of (start, end) indices for each chain, where end is not inclusive.

Iterates over the residues in the structure.

Runs MSA and template tools and returns a new Input with the results.

Queries the database using hmmsearch using a given hmm.

Computes a forward pass of the model on a featurised example.

Sharded apply. Applies `fun` over shards to axes, in a way similar to vmap, but does so in shards of `shard_size`. Shards are stacked after. Th

Connects Module. Args: inputs: Tensor of shape [..., num_channel] Returns: output of shape [..., num_output]

Gets raw field data using field name as a string.

Constructs Linear Module. Args: num_output: number of output channels. Can be tuple when outputting multiple dimensions. init

Adds a ligand to chemical components. Raises ValueError on mismatch.

Aligns MSA row indices based on species. Within a species, MSAs are aligned based on their original order (the first sequence for a species in th

Applies a bioassembly transform to this structure.

Creates a sequence of `Structure` objects with unique chain IDs. Let e.g. [A, B] denote a structure of two chains A and B, then this function per

Compares chirality of two Mols. Mol1 can match a subset of mol2.

Compute asym_id, entity_id and sym_id. Args: all_tokens: atom layout containing a representative atom for each token. Returns: A Chains

Computes the chain pair ipTM metrics from PAE.

Compute covariance matrix. The quantity computes is cov_xy = weighted_avg_i(row_values[i, x] col_values[j, y]). Here x and y run over the xyz c

Creates maps to help find bonds during a loop over residues.

Embeds bond features and merges into pair activations.

Processes MSA and returns updated pair activations.

Embeds Templates and merges into pair activations.

Encodes a sequence of residue names as a sequence of ints. Args: chain_poly_type: Polymer chain type to determine sequence encoding. sequen

Extracts sequence identifier from description. Returns None if no match.

Formats an atom name to fit in the four characters specified in PDB. See for example the following note on atom name formatting in PDB files: htt

Generates all required tables and columns if missing.

Returns chain key for each chain ID respecting resolved chain ordering.

Returns first definitely non-leaving atom if exists, as a stand-in.

Updates a drop_leaving_atoms mask with new leaving atom locations.

Returns features for the ligands in this structure.

Extract header fields from an mmCIF object.

Parses an `Operation` from a mmCIF _pdbx_struct_oper_list row.

Processes a single protein chain.

Creates reference positions from a CCD mmcif data block.

Processes a single RNA chain.

Get SMILES string from chemical components.

Converts a user-specified model_id argument into a string.

Generate a grid of rng keys that is consistent with different padding. Generate random keys such that the keys will be identical, regardless of h

Guess the entity type (polymer/non-polymer/water) based on residues/atoms. We treat both arguments as unordered collections since we care only whet

Loads the CCD pickle file and caches it so that it is only loaded once.

Map from dense_atom to atom37 per residue type.

Returns indices of residue's pseudo-beta.

Create Mapping from rigid_groups to dense_atom indices.

Match files in a directory with a pattern, and group by model name.