Using featurizers

In [1]:

import logging
from typing import List, Optional
import pandas as pd

from polymetrix.featurizers.polymer import Polymer
from polymetrix.featurizers.chemical_featurizer import (
    NumHBondDonors,
    NumHBondAcceptors,
    NumRotatableBonds,
    NumRings,
    NumNonAromaticRings,
    NumAromaticRings,
    NumAtoms,
    TopologicalSurfaceArea,
    FractionBicyclicRings,
    NumAliphaticHeterocycles,
    SlogPVSA1,
    BalabanJIndex,
    MolecularWeight,
    Sp3CarbonCountFeaturizer,
    Sp2CarbonCountFeaturizer,
    MaxEStateIndex,
    SmrVSA5,
    FpDensityMorgan1,
    HalogenCounts,
    BondCounts,
    BridgingRingsCount,
    MaxRingSize,
    HeteroatomCount,
    HeteroatomDensity,
)
from polymetrix.featurizers.sidechain_backbone_featurizer import (
    SideChainFeaturizer,
    NumSideChainFeaturizer,
    BackBoneFeaturizer,
    NumBackBoneFeaturizer,
    FullPolymerFeaturizer,
    SidechainLengthToStarAttachmentDistanceRatioFeaturizer,
    StarToSidechainMinDistanceFeaturizer,
    SidechainDiversityFeaturizer,
)
from polymetrix.featurizers.multiple_featurizer import MultipleFeaturizer

import logging from typing import List, Optional import pandas as pd from polymetrix.featurizers.polymer import Polymer from polymetrix.featurizers.chemical_featurizer import ( NumHBondDonors, NumHBondAcceptors, NumRotatableBonds, NumRings, NumNonAromaticRings, NumAromaticRings, NumAtoms, TopologicalSurfaceArea, FractionBicyclicRings, NumAliphaticHeterocycles, SlogPVSA1, BalabanJIndex, MolecularWeight, Sp3CarbonCountFeaturizer, Sp2CarbonCountFeaturizer, MaxEStateIndex, SmrVSA5, FpDensityMorgan1, HalogenCounts, BondCounts, BridgingRingsCount, MaxRingSize, HeteroatomCount, HeteroatomDensity, ) from polymetrix.featurizers.sidechain_backbone_featurizer import ( SideChainFeaturizer, NumSideChainFeaturizer, BackBoneFeaturizer, NumBackBoneFeaturizer, FullPolymerFeaturizer, SidechainLengthToStarAttachmentDistanceRatioFeaturizer, StarToSidechainMinDistanceFeaturizer, SidechainDiversityFeaturizer, ) from polymetrix.featurizers.multiple_featurizer import MultipleFeaturizer

Full Polymer Featurization¶

In [2]:

psmiles_list = [
    "c1ccccc1[*]CCO[*]",
    "CC[*]CCCC[*]",
]

full_featurizers = [
    FullPolymerFeaturizer(NumRings()),
    FullPolymerFeaturizer(MolecularWeight()),
    FullPolymerFeaturizer(TopologicalSurfaceArea()),
]
full_multi_featurizer = MultipleFeaturizer(full_featurizers)

for psmiles in psmiles_list:
    polymer = Polymer.from_psmiles(psmiles)
    features = full_multi_featurizer.featurize(polymer)
    labels = full_multi_featurizer.feature_labels()

    for label, value in zip(labels, features):
        print(f"{label}: {value:.2f}")

psmiles_list = [ "c1ccccc1[*]CCO[*]", "CC[*]CCCC[*]", ] full_featurizers = [ FullPolymerFeaturizer(NumRings()), FullPolymerFeaturizer(MolecularWeight()), FullPolymerFeaturizer(TopologicalSurfaceArea()), ] full_multi_featurizer = MultipleFeaturizer(full_featurizers) for psmiles in psmiles_list: polymer = Polymer.from_psmiles(psmiles) features = full_multi_featurizer.featurize(polymer) labels = full_multi_featurizer.feature_labels() for label, value in zip(labels, features): print(f"{label}: {value:.2f}")

num_rings_sum_fullpolymerfeaturizer: 1.00
molecular_weight_sum_fullpolymerfeaturizer: 121.07
topological_surface_area_sum_fullpolymerfeaturizer: 9.23
num_rings_sum_fullpolymerfeaturizer: 0.00
molecular_weight_sum_fullpolymerfeaturizer: 85.10
topological_surface_area_sum_fullpolymerfeaturizer: 0.00

Side Chain Featurization¶

In [3]:

sidechain_featurizers = [
    NumSideChainFeaturizer(),
    SideChainFeaturizer(NumAtoms(agg=["sum"])),
    SideChainFeaturizer(NumHBondDonors(agg=["sum"])),
    SideChainFeaturizer(NumRotatableBonds(agg=["sum"])),
]
sidechain_multi_featurizer = MultipleFeaturizer(sidechain_featurizers)

for psmiles in psmiles_list:
    polymer = Polymer.from_psmiles(psmiles)
    features = sidechain_multi_featurizer.featurize(polymer)
    labels = sidechain_multi_featurizer.feature_labels()
    
    for label, value in zip(labels, features):
        print(f"{label}: {value:.2f}")

sidechain_featurizers = [ NumSideChainFeaturizer(), SideChainFeaturizer(NumAtoms(agg=["sum"])), SideChainFeaturizer(NumHBondDonors(agg=["sum"])), SideChainFeaturizer(NumRotatableBonds(agg=["sum"])), ] sidechain_multi_featurizer = MultipleFeaturizer(sidechain_featurizers) for psmiles in psmiles_list: polymer = Polymer.from_psmiles(psmiles) features = sidechain_multi_featurizer.featurize(polymer) labels = sidechain_multi_featurizer.feature_labels() for label, value in zip(labels, features): print(f"{label}: {value:.2f}")

numsidechainfeaturizer: 1.00
num_atoms_sidechainfeaturizer_sum: 6.00
num_hbond_donors_sidechainfeaturizer_sum: 0.00
num_rotatable_bonds_sidechainfeaturizer_sum: 0.00
numsidechainfeaturizer: 1.00
num_atoms_sidechainfeaturizer_sum: 2.00
num_hbond_donors_sidechainfeaturizer_sum: 0.00
num_rotatable_bonds_sidechainfeaturizer_sum: 0.00

Backbone Featurization¶

In [4]:

backbone_featurizers = [
    NumBackBoneFeaturizer(),
    BackBoneFeaturizer(NumRings()),
    BackBoneFeaturizer(NumAtoms()),
    BackBoneFeaturizer(TopologicalSurfaceArea()),
]
backbone_multi_featurizer = MultipleFeaturizer(backbone_featurizers)

for psmiles in psmiles_list:
    polymer = Polymer.from_psmiles(psmiles)
    features = backbone_multi_featurizer.featurize(polymer)
    labels = backbone_multi_featurizer.feature_labels()
    
    for label, value in zip(labels, features):
        print(f"{label}: {value:.2f}")

backbone_featurizers = [ NumBackBoneFeaturizer(), BackBoneFeaturizer(NumRings()), BackBoneFeaturizer(NumAtoms()), BackBoneFeaturizer(TopologicalSurfaceArea()), ] backbone_multi_featurizer = MultipleFeaturizer(backbone_featurizers) for psmiles in psmiles_list: polymer = Polymer.from_psmiles(psmiles) features = backbone_multi_featurizer.featurize(polymer) labels = backbone_multi_featurizer.feature_labels() for label, value in zip(labels, features): print(f"{label}: {value:.2f}")

numbackbonefeaturizer: 1.00
num_rings_sum_backbonefeaturizer: 0.00
num_atoms_sum_backbonefeaturizer: 5.00
topological_surface_area_sum_backbonefeaturizer: 9.23
numbackbonefeaturizer: 1.00
num_rings_sum_backbonefeaturizer: 0.00
num_atoms_sum_backbonefeaturizer: 6.00
topological_surface_area_sum_backbonefeaturizer: 0.00