If missing, infers the scheme tables from the _atom_site table.

Creates a Mol object from CCD information if res_name is in the CCD.

Merges Jackhmmer result protos into a single one.

Merges nhmmer result protos into a single one.

Loads model parameters and returns a jitted model forward pass.

Creates a Mol object from a ligand structure and reference mol.

Pad a set of features to the maximum size amongst all chains. Args: feat_name: The feature name to pad. chains: A list of chains with assoc

Modify any element-wise random function to be consistent with padding. Normally if you take a function like jax.random.normal and generate an array

Returns the bonds table extracted from the mmCIF. Args: cif: The raw mmCIF to extract the bond information from. atom_key: A numpy array de

Parses the hmmsearch A3M sequence description line.

Parse hit metadata by parsing mmCIF from structure store.

Builds a conformer containing the ideal coordinates from the CCD. Args: mol_cif: An mmcif object representing a molecule. Returns: An

Gets species from an msa sequence identifier. The sequence identifier has the format specified by _UNIPROT_TREMBL_ENTRY_NAME_PATTERN or _UNIPROT_

computes single and pair conditioning for all atoms in each token.

Populate the atoms of a Mol given atom features. Args: mol: Mol object. atom_names: Names of the atoms. atom_types: Types of the atoms.

Populate the bonds of a Mol given bond features. Args: mol: Mol object. atom_names: Names of atoms in the molecule. bond_begins: Names

Reads a record encoded by `_encode_record` from a byte stream.

Reduces template features to max num templates and defined feature set.

Add relative position encodings.

Remaps the query -> hit mapping to match the actual Structure sequence. Args: hit_seq: The hit sequence - a subsequence of the PDB seqres seque

Remove instances greedily.

Returns record names (ATOM/HETATM) given residue names and polymer mask.

Resolves path for data dep paths, stringifies otherwise.

Generated Pair embedding from sequence.

Transforms type annotations of the form `T | None` to `T`.

Unpacks filter kwargs into predicates for each table.

Validates that all tables have the same ordering.

Validates that all foreign keys are present in the referred tables.

Validates the keys of the user-defined CCD dictionary.

Returns a structure with new bonds added. Args: bonded_atom_pairs: A sequence of pairs of atoms, with one pair per bond. Each eleme

Align x to y considering only included_idxs. Args: x: NxD np array of coordinates. y: NxD np array of coordinates. x_indices: An np arr

Aligns the sequences to profile and returns the alignment in A3M string. Uses hmmalign to align the sequences to the profile, then ouputs the

Returns a sliced table using a key (!= index) array or a boolean mask.

Walks all residues, making sure alll elements are equal.

Walks naming scheme, making sure all elements are equal.

Checks whether two Bonds objects are considered equal.

Checks whether two ChemicalComponentData objects are considered equal.

Assigns atom names from the molecular graph. The atom name is stored as an atom property 'atom_name', accessible with atom.GetProp('atom_name').

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

Calculates the bucket size to pad the data to.

Compute predicted TM (pTM) between each pair of chains independently. Args: tm_adjusted_pae: [num_res, num_res] Relevant tensor for computing T

Produce gather indices and mask to convert from source layout to target.

Performs a shallow copy but with coordinates updated.

Create target feature embedding.

Multihead self-attention.

Returns the raw per-atom deviations used in RMSD computation.

Compute distogram from amino acid positions. Args: positions: (num_res, 3) Position coordinates. config: Distogram bin configuration. Re

Drops atoms that are not in the CCD for the given residue type.

Creates a fully masked out template features to allow padding to work. Args: num_res: The length of the target chain. Returns: Empty tem

Returns full residue names based on a sequence string. Args: sequence: A string representing the sequence. chain_type: The chain type of th

Checks whether this ChemCompEntry extends another one.

Extracts distogram from model outputs.

Returns a sequence of chain IDs that contain only unknown residues.

Finds glycans and ions in all version of CCD.

Remaps residue names to the closest standard protein/RNA/DNA residue. If residue name is already a standard type, it is not altered. If a match c

Constructs Input from an AlphaFoldServer fold job.

Construct Rot3Array Matrix from array of shape. [..., 3, 3].

Creates Templates from a Hmmsearch A3M. Args: query_sequence: The polymer sequence of the target query. a3m: Results of Hmmsearch in

Initializes the MSA from multiple MSAs. Args: msas: A sequence of Msa objects representing individual MSAs produced by different to

Construct a Structure from a parsed mmCIF object. This function is called by `from_mmcif` but can be useful when an mmCIF has already been parsed

Creates templates from a run of hmmsearch tool against a custom a3m. Args: query_sequence: The polymer sequence of the target query.

Constructs Rot3Array from arbitrary array of shape [3 * 3] using SVD. The case when 'use_quat_formula' is False rephrases the problem of proj

Construct Rot3Array from two Vectors. Rot3Array is constructed such that in the corresponding frame 'e0' lies on the positive x-Axis and 'e1'

Generates a biological assembly as a new `Structure`. When no assembly ID is provided this method produces a default assembly. If this struct

Get all possible atoms and bonds for this residue in a standard order. Args: ccd: The chemical components dictionary. res_name: Full CCD na

Finds the res_name on the opposite end of the bond, if a bond exists. Args: polymer_ligand_bonds: Bond information for polymer-ligand pairs.

Returns `ChemicalComponentsData` for chemical components known by PDB.

Gets a default assembly ID.

Get the predicted structure, scalars, and arrays for inference. This function also computes any inference-time quantities, which are not a pa

Returns set of atoms that are dropped when this res_name gets linked. Args: res_name: residue name, e.g. 'ARG' chain_type: chain_type, e.g.

Returns the requested MSA tool.

Returns features for this polymer chain. Args: chain: Structure object representing the template. Must be already filtered to a single ch

Creates the predicted structure and ion preditions. Args: result: model output in a model specific layout batch: model input batch Retur

Reference structure for residue from CCD or SMILES. Uses CCD entry if available, otherwise uses SMILES from chemical components data. Conformer g

Returns chain, residue, and atom tables from a parsed mmcif. Args: cif: A parsed mmcif.Mmcif. fix_mse_residues: See from_mmcif. fix_arg

Gets the value of a column at the row with specified key value.

Samples with replacement from a distribution given by 'logits'. This uses Gumbel trick to implement the sampling an efficient manner. For a distr

Returns whether the hit should be kept. In addition to filtering on all of the provided parameters, this method also excludes hits with unres

Loads multiple fold inputs from a JSON string.

Make backbone Rigid3Array and mask. Args: positions: (num_res, num_atoms) of atom positions as Vec3Array. mask: (num_res, num_atoms) for at

Returns the Structure chains table.

Generates factors for converting 3x3 matrix to symmetric 4x4 matrix.