Adding data into HIPPO

In theory the HIPPO class makes it simple to insert new material into the database, but there are nuances, especially to improve performance.

Registering compounds

The HIPPO.register_compound() method takes as a minimum a smiles string and adds a compound to the database:

c = animal.register_compound(smiles='Clc1c[nH]cn1')

N.B. to prevent accidents, positional arguments are disabled for many methods in HIPPO.

In the case that a compound with this structure already exists, the compound is instead retrieved from the database (duplicates will never be created).

Tags, bases, and metadata

Additional detail can be added to the compound upon registration:

• You may wish to tag the compound with any number of strings.

• You can assign a ‘base’ compound of which this compound is a superstructure.

• Any dictionary of additional metadata can be stored

c = animal.register_compound(
        smiles='Clc1c[nH]cn1',
        tags=['fragment', 'hit', 'halogen'],
        metadata={'screening_library':'dsiPoised'},
)

See also the API reference HIPPO.register_compound().

Registering a Target and associated Poses

The Target class currently only stores a name, but it is a requirement that each Pose is associated with a single Target.

To create (or retrieve) a target:

a71 = animal.register_target(name="A71EV2A")

Then a pose can be registered:

animal.register_pose(
        compound=c,
        target=a71,
        path=/path/to/bound.pdb, # protein ligand complex PDB
)

Optional arguments include:

• tags

• metadata

• inspirations (other poses that inspired this one)

• reference (in case your path is to a file containing only a ligand .mol, the reference points to another pose who’s protein will be used)

See also the API reference HIPPO.register_target() and HIPPO.register_pose().

Registering a reaction

To add a reaction:

animal.register_reaction(
        type="Amidation",
        product=c,
        reactants=[a,b],
)

The above will register an amidation reaction combining compounds a and b into c.

See also the API reference HIPPO.register_reaction().

Improving Performance

Querying a large database can become expensive, and permance gains can also be realised by commiting database transactions in bulk. Additionally, initialising python object instances like Compound may not be necessary.

In a large loop of registering compounds, reactions, and poses the above issues can be mitigated:

for i,row in df.iterrows():

        # register the reactants

        reactant1 = animal.register_compound(
                smiles=row.r1_smiles,
                return_compound=False, # this just returns the compound ID
                commit=False, # this does not commit the change to the database
        )

        reactant2 = animal.register_compound(
                smiles=row.r2_smiles,
                return_compound=False, # this just returns the compound ID
                commit=False, # this does not commit the change to the database
        )

        # register the product

        product = animal.register_compound(
                smiles=row.smiles,
                return_compound=True,
                metadata=comp_metadata,
                commit=False
        )

        # register the reaction

        reaction = animal.register_reaction(
                type=row.reaction,
                product=product,
                reactants=[reactant1, reactant2],
                commit=False,
        )

        # register the pose

        pose = animal.register_pose(
                compound=product,
                target='A71EV2A',
                path=pose_path,
                metadata=pose_metadata,
                inspirations=[inspiration1, inspiration2],
                commit=False,
                return_pose=False,
                overwrite_metadata=True, # don't bother checking existing metadata
        )

        # add tags and bases:

        animal.db.update(
                table='compound',
                id=product_id,
                key='compound_base',
                value=base_id,
                commit=False
        )

        animal.db.insert_tag(
                name='elab',
                compound=product_id,
                commit=False,
        )

        animal.db.commit()