try:
  import google.colab
  %pip install -q ugropy
except ImportError:
  pass

Gibbs Models

In some situations, you may not need to instantiate all the models supported by ugropy. Instead, you can search for a specific model’s groups individually.

Warning

For the UNIFAC, PSRK, and Dortmund models, the aldehyde group name has been changed to HCO, in accordance with the following discussion:

https://github.com/ClapeyronThermo/Clapeyron.jl/issues/225

This is more consistent with the ether groups and formate group.

UNIFAC

The classic liquid-vapor UNIFAC model can be imported and used by doing:

from ugropy import unifac

hexane = unifac.get_groups("n-hexane")

print(hexane.subgroups)

{'CH3': 2, 'CH2': 4}

hexane.draw()

As you can see, the UNIFAC model can be used just like we did on the Groups class. The Groups class is only intended to be used when you want all the groups from all the models with a single call.

The signature of the get_groups method is:

unifac.get_groups(
    identifier: Union[str, Chem.rdchem.Mol],
    identifier_type: str = "name",
    solver: ILPSolver = DefaultSolver,
    search_multiple_solutions: bool = False,
    search_nonoptimal: bool = False,
) -> Union[GibbsFragmentationResult, List[GibbsFragmentationResult]]:
    """Get the groups of a molecule.

    Parameters
    ----------
    identifier : Union[str, Chem.rdchem.Mol]
        Identifier of the molecule. You can use either the name of the
        molecule, the SMILEs of the molecule or a rdkit Mol object.
    identifier_type : str, optional
        Identifier type of the molecule. Use "name" if you are providing
        the molecules' name, "smiles" if you are providing the SMILES
        or "mol" if you are providing a rdkir mol object, by default "name"
    solver : ILPSolver, optional
        ILP solver class, by default DefaultSolver
    search_multiple_solutions : bool, optional
        Weather search for multiple solutions or not, by default False
        If False the return will be a FragmentationResult object, if True
        the return will be a list of FragmentationResult objects.
    search_nonoptimal : bool, optional
        If True, the solver will search for non-optimal solutions along
        with the optimal ones. This is useful when the user wants to find
        all possible combinations of fragments that cover the universe. By
        default False. If `search_multiple_solutions` is False, this
        parameter will be ignored.

    Returns
    -------
    Union[GibbsFragmentationResult, List[GibbsFragmentationResult]]
        Fragmentation result. If search_multiple_solutions is False the
        return will be a FragmentationResult object, if True the return
        will be a list of FragmentationResult objects.
    """

As you can see, it has the same arguments as the Groups class.

Let’s obtain multiple solution by doing:

mol = unifac.get_groups("CCCC1=CC=C(CC(=O)OC)C=C1", "smiles", search_multiple_solutions=True)

mol

[<ugropy.core.frag_classes.gibbs_model.gibbs_result.GibbsFragmentationResult at 0x7f45985d6120>,
 <ugropy.core.frag_classes.gibbs_model.gibbs_result.GibbsFragmentationResult at 0x7f4557f308c0>]

mol[0].subgroups

{'CH3': 2, 'CH2': 1, 'ACH': 4, 'ACCH2': 2, 'COO': 1}

mol[1].subgroups

{'CH3': 2, 'CH2': 1, 'ACH': 4, 'AC': 1, 'ACCH2': 1, 'CH2COO': 1}

mol[0].draw(width=800)

mol[1].draw(width=800)

Finally GibbsModels can estimate the \(R\) and \(Q\) (volume and surface area) of a molecule, very useful for the UNIQUAC model.

chex = unifac.get_groups("cyclohexane")

chex.r, chex.q

(np.float64(4.0464), np.float64(3.24))

chex.draw()

PSRK

The Predictive Soave-Redlich-Kwong model can be imported and used by doing:

from ugropy import psrk

toluene = psrk.get_groups("toluene")

toluene.subgroups

{'ACH': 5, 'ACCH3': 1}

PSRK model works the same as the UNIFAC model.

toluene.draw()

toluene.r, toluene.q

(np.float64(3.9227999999999996), np.float64(2.968))

Dortmund (modified UNIFAC)

The Dortmund model can be imported and used by doing:

from ugropy import dortmund


cyclohexane = dortmund.get_groups("cyclohexane")

cyclohexane.subgroups

{'CY-CH2': 6}

cyclohexane.draw()

Since the \(R\) and \(Q\) values of the Dortmund model are fitted along the interaction parameters and do not represent the volume and surface area of the molecule, the values are not calculated by ugropy to avoid confusion.

print(cyclohexane.r, cyclohexane.q)

None None