def _parse_residue()

in mmcif_utils.py [0:0]

• 33 lines of code
• 8 McCabe index (conditional complexity)

def _parse_residue(residue):
    """Obtains a sparse representation of residue from gemmi"""

    # list of atoms in the residue (e.g. N-CA-C-O)
    atoms = res_atoms[residue.name]

    # ordinal encoding of how far away the parents are
    parents = res_parents[residue.name]

    # ordinal encoding of how far away the children are
    children = res_children[residue.name]

    # atoms belonging to chi anglse
    chis = res_chis[residue.name]

    # accumulate the xyz postions of the atoms, and node_embed encodings
    pos, node_embeds = [], []

    residue_counter = 0
    for atom in atoms:
        if atom in residue:
            atom = residue[atom]
        elif recorrect_name(atom) in residue:
            atom = residue[recorrect_name(atom)]
        else:
            return None

        pos.append((atom.pos.x, atom.pos.y, atom.pos.z))
        node_embeds.append(
            (
                residue_names.index(residue.name),
                atom_names.index(atom.element.name),
                residue_counter,
                atom.pos.x,
                atom.pos.y,
                atom.pos.z,
            )
        )
        residue_counter = residue_counter + 1

    # 20-dim mask for each residue for atom existence
    exist = [1] * len(parents) + [0] * (20 - len(parents))

    # pad the parents and children to 20-dim
    parents = parents + [0] * (20 - len(parents))
    children = children + [0] * (20 - len(children))

    # place the x,y,z coordinates into a numpy array
    pos_fill = np.zeros((20, 3))
    pos_fill[: len(pos)] = pos

    # pad the chi angles
    chis = [1] * len(chis) + [0] * (5 - len(chis))

    # return the new representation
    return parents, children, pos_fill, exist, chis, node_embeds