Returns a model config with some defaults overridden.

Returns the Structure residues table.

Converts CCD mmCIFs to component info. Missing fields are left empty.

Returns a new structure with atoms ordered & dropped to match another's. This performs two operations simultaneously: * Ordering the atoms

Stacks polymer features, adds empty and keeps ligand features unstacked.

Parses atoms. If force_parse is True, fix deuterium and missing charge.

Parses _chem_comp_atom data from an Mmcif object.

Parses bond data. If force_parse is True, ignore missing aromatic flags.

Parses _chem_comp_bond data from an Mmcif object.

Parses FASTA string and returns list of strings with amino-acid sequences. Arguments: fasta_string: The string contents of a FASTA file. Ret

Returns the shard spec or None if the path is not a shard spec. For instance, if the shard spec is '/path/to/file@20', the output will be ('/path

Mapping from author chain ID to single letter aa sequence, polymers only. This mapping is well defined only for polymers (protein, DNA, RNA), but

Returns cif, confidence_1d_json, confidence_2d_json, mean_confidence_1d, and ranking confidence.

Runs the full inference pipeline to predict structures for each seed.

Computes predicted TM alignment or predicted interface TM alignment score. Args: tm_adjusted_pae: [num_res, num_res] Relevant tensor for comput

Takes requests from in_queue, adds (key, serialized ex) to out_queue.

Processes a single protein chain.

Processes a single RNA chain.

Computes features for a structure and associated MSAs/templates.

Query the database (sharded or unsharded) using Nhmmer.

Queries the database using hmmsearch using a given stockholm msa.

Center then apply random translation and rotation.

Apply random rigid augmentation. Args: rng_key: random key positions: atom positions of shape (<common_axes>, 3) mask: per-atom mask of

Fully reads the contents of a byte stream.

Compute 1D summaries from a chain-pair summary. Args: chain_pair_met: A [num_samples, num_chains, num_chains] aggregate matrix. num_chain_t

Returns a new BioassemblyData with renamed label_asym_ids. Args: mapping: A mapping from original label_asym_ids to their new values. Any

Reorders tables so that the label_asym_ids are in the given order. This method changes the order of the chains, residues, and atoms tables so

Replaces the DB_DIR placeholder in a path with the given DB_DIR.

Runs Hmmsearch to get a3m string of hits.

Select the model files from a model directory.

Multihead self-attention.

Sorts the atoms in the molecule by their names.

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

Yields the names of the tables in the database in dependency order. This order guarantees that a table appears after all other tables that it ref

Stacks the fields to a numpy array with shape (6, <layout_shape>). Creates a pure numpy array of type `object` by stacking the 6 fields of the

Converts to a sequence of CCD codes.

Converts to a sequence of CCD codes.

Converts Input to an AlphaFold JSON.

Returns a dict suitable for building a CifDict, representing bonds. Args: atom_key: A (num_atom,) integer array of atom_keys. chain_i

Returns atom-level information in arrays containing a num_res dimension. NB: All residues in the structure will appear in the residue dimension b

Converts Input to a Structure. WARNING: This method does not preserve the rng seeds. Args: ccd: The chemical components dictionary.

Find the least squares best fit rotation between two sets of N points. Solve Ax = b for A. Where A is the transform rotating x^T into b^T. Args:

Validates the fold input contains MSA and templates for featurisation.

Implementation of AlphaFold-RSA. AlphaFold-RSA defined in https://www.ncbi.nlm.nih.gov/pmc/articles/PMC9601767. Args: cif: Raw cif string.

Returns a copy of the input with num_seeds rng seeds.

Returns the structure without hydrogen atoms.

Returns the structure without terminal oxygen atoms.

Writes the input JSON to the output directory.

Writes outputs to the specified output directory.

Builds a module. Arguments: act: [num_seq, num_res, channels] activations tensor pair_mask: [num_seq, num_res] mask of non-padded reg

Applies Module. Args: act: The activation. mask: The mask. Returns: Outputs, should have same shape/type as output_act

Build a single iteration of the pair former. Args: act: [num_res, num_res, num_channel] Input pairwise activations. pair_mask: [num_r

Builds ConfidenceHead module. Arguments: dense_atom_positions: [N_res, N_atom, 3] array of positions. embeddings: Dictionary of repre

Generate an embedding for a set of templates. Args: query_embedding: [num_res, num_res, num_channel] a query tensor that will be us

Build the single template embedding graph. Args: query_embedding: (num_res, num_res, num_channels) - embedding of the query sequenc

Pickle calls this on dump. Returns: Members with cached properties removed.

Returns members with cached properties removed for pickling.