Using featurizers

In [1]:

%reload_ext autoreload
%autoreload 2

%reload_ext autoreload %autoreload 2

In [2]:

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

from polymetrix.featurizers.polymer import Polymer
from polymetrix.featurizers.molecule import Molecule
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.molecule import FullMolecularFeaturizer

from polymetrix.featurizers.multiple_featurizer import MultipleFeaturizer

from polymetrix.featurizers.comparator import PolymerMoleculeComparator

import logging from typing import List, Optional import pandas as pd from polymetrix.featurizers.polymer import Polymer from polymetrix.featurizers.molecule import Molecule 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.molecule import FullMolecularFeaturizer from polymetrix.featurizers.multiple_featurizer import MultipleFeaturizer from polymetrix.featurizers.comparator import PolymerMoleculeComparator

Full Polymer Featurization¶

In [3]:

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 [4]:

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 [5]:

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

Full Molecular Featurization¶

In [6]:

psmiles_list = [
    "CCCC",
    "NC(=O)c1ccc2c(c1)nc(C1CCC(O)CC1)n2CCCO",
    "CNC(=S)Nc1cccc(-c2cnc3ccccc3n2)c1",
    "C#Cc1ccc(-c2nc(-c3cc[nH]c(=O)c3)c(-c3ccc(F)cc3)[nH]2)cc1",
]

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

for psmiles in psmiles_list:
    polymer = Molecule.from_smiles(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 = [ "CCCC", "NC(=O)c1ccc2c(c1)nc(C1CCC(O)CC1)n2CCCO", "CNC(=S)Nc1cccc(-c2cnc3ccccc3n2)c1", "C#Cc1ccc(-c2nc(-c3cc[nH]c(=O)c3)c(-c3ccc(F)cc3)[nH]2)cc1", ] full_featurizers = [ FullMolecularFeaturizer(NumRings()), FullMolecularFeaturizer(MolecularWeight()), FullMolecularFeaturizer(TopologicalSurfaceArea()), ] full_multi_featurizer = MultipleFeaturizer(full_featurizers) for psmiles in psmiles_list: polymer = Molecule.from_smiles(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_fullmolecularfeaturizer: 0.00
molecular_weight_sum_fullmolecularfeaturizer: 58.08
topological_surface_area_sum_fullmolecularfeaturizer: 0.00
num_rings_sum_fullmolecularfeaturizer: 3.00
molecular_weight_sum_fullmolecularfeaturizer: 317.17
topological_surface_area_sum_fullmolecularfeaturizer: 101.37
num_rings_sum_fullmolecularfeaturizer: 3.00
molecular_weight_sum_fullmolecularfeaturizer: 294.09
topological_surface_area_sum_fullmolecularfeaturizer: 49.84
num_rings_sum_fullmolecularfeaturizer: 4.00
molecular_weight_sum_fullmolecularfeaturizer: 355.11
topological_surface_area_sum_fullmolecularfeaturizer: 61.54

Using Comparators to Compare Polymer and Molecule Features¶

In [7]:

polymer = Polymer.from_psmiles('*CCCCCCNC(=O)c1ccc(C(=O)N*)c(Sc2ccccc2)c1')
molecule = Molecule.from_smiles('CC(=O)OC1=CC=CC=C1C(=O)O')

polymer_featurizers = [
    FullPolymerFeaturizer(MolecularWeight()),
    FullPolymerFeaturizer(NumHBondDonors()),
    FullPolymerFeaturizer(NumHBondAcceptors()),
    FullPolymerFeaturizer(NumRotatableBonds())
]

molecule_featurizers = [
    FullMolecularFeaturizer(MolecularWeight()),
    FullMolecularFeaturizer(NumHBondDonors()),
    FullMolecularFeaturizer(NumHBondAcceptors()),
    FullMolecularFeaturizer(NumRotatableBonds())
]

polymer_multi = MultipleFeaturizer(polymer_featurizers)
molecule_multi = MultipleFeaturizer(molecule_featurizers)

# comparator = PolymerMoleculeComparator(polymer_multi, molecule_multi)

comparator = PolymerMoleculeComparator(
    polymer_multi, 
    molecule_multi, 
    comparisons=["absolute_difference", "signed_difference", "product", "squared_distance", "euclidean_distance"],
    agg=["mean", "max", "min", "sum"]
)

difference = comparator.compare(polymer, molecule)

labels = comparator.feature_labels()

# Print feature-wise results
for label, diff in zip(labels, difference):
    print(f"  {label}: {diff}")

polymer = Polymer.from_psmiles('*CCCCCCNC(=O)c1ccc(C(=O)N*)c(Sc2ccccc2)c1') molecule = Molecule.from_smiles('CC(=O)OC1=CC=CC=C1C(=O)O') polymer_featurizers = [ FullPolymerFeaturizer(MolecularWeight()), FullPolymerFeaturizer(NumHBondDonors()), FullPolymerFeaturizer(NumHBondAcceptors()), FullPolymerFeaturizer(NumRotatableBonds()) ] molecule_featurizers = [ FullMolecularFeaturizer(MolecularWeight()), FullMolecularFeaturizer(NumHBondDonors()), FullMolecularFeaturizer(NumHBondAcceptors()), FullMolecularFeaturizer(NumRotatableBonds()) ] polymer_multi = MultipleFeaturizer(polymer_featurizers) molecule_multi = MultipleFeaturizer(molecule_featurizers) # comparator = PolymerMoleculeComparator(polymer_multi, molecule_multi) comparator = PolymerMoleculeComparator( polymer_multi, molecule_multi, comparisons=["absolute_difference", "signed_difference", "product", "squared_distance", "euclidean_distance"], agg=["mean", "max", "min", "sum"] ) difference = comparator.compare(polymer, molecule) labels = comparator.feature_labels() # Print feature-wise results for label, diff in zip(labels, difference): print(f" {label}: {diff}")

  molecular_weight_sum_fullpolymerfeaturizer_absolute_difference: 174.097940208
  num_hbond_donors_sum_fullpolymerfeaturizer_absolute_difference: 1.0
  num_hbond_acceptors_sum_fullpolymerfeaturizer_absolute_difference: 0.0
  num_rotatable_bonds_sum_fullpolymerfeaturizer_absolute_difference: 8.0
  molecular_weight_sum_fullpolymerfeaturizer_signed_difference: 174.097940208
  num_hbond_donors_sum_fullpolymerfeaturizer_signed_difference: 1.0
  num_hbond_acceptors_sum_fullpolymerfeaturizer_signed_difference: 0.0
  num_rotatable_bonds_sum_fullpolymerfeaturizer_signed_difference: 8.0
  molecular_weight_sum_fullpolymerfeaturizer_product: 63760.20132709417
  num_hbond_donors_sum_fullpolymerfeaturizer_product: 2.0
  num_hbond_acceptors_sum_fullpolymerfeaturizer_product: 9.0
  num_rotatable_bonds_sum_fullpolymerfeaturizer_product: 20.0
  molecular_weight_sum_fullpolymerfeaturizer_squared_distance: 30310.092784668348
  num_hbond_donors_sum_fullpolymerfeaturizer_squared_distance: 1.0
  num_hbond_acceptors_sum_fullpolymerfeaturizer_squared_distance: 0.0
  num_rotatable_bonds_sum_fullpolymerfeaturizer_squared_distance: 64.0
  molecular_weight_sum_fullpolymerfeaturizer_euclidean_distance: 174.097940208
  num_hbond_donors_sum_fullpolymerfeaturizer_euclidean_distance: 1.0
  num_hbond_acceptors_sum_fullpolymerfeaturizer_euclidean_distance: 0.0
  num_rotatable_bonds_sum_fullpolymerfeaturizer_euclidean_distance: 8.0
  molecular_weight_sum_fullpolymerfeaturizer_absolute_difference_signed_difference_product_squared_distance_euclidean_distance_mean: 18918.5175864773
  num_hbond_donors_sum_fullpolymerfeaturizer_absolute_difference_signed_difference_product_squared_distance_euclidean_distance_mean: 1.2
  num_hbond_acceptors_sum_fullpolymerfeaturizer_absolute_difference_signed_difference_product_squared_distance_euclidean_distance_mean: 1.8
  num_rotatable_bonds_sum_fullpolymerfeaturizer_absolute_difference_signed_difference_product_squared_distance_euclidean_distance_mean: 21.6
  molecular_weight_sum_fullpolymerfeaturizer_absolute_difference_signed_difference_product_squared_distance_euclidean_distance_max: 63760.20132709417
  num_hbond_donors_sum_fullpolymerfeaturizer_absolute_difference_signed_difference_product_squared_distance_euclidean_distance_max: 2.0
  num_hbond_acceptors_sum_fullpolymerfeaturizer_absolute_difference_signed_difference_product_squared_distance_euclidean_distance_max: 9.0
  num_rotatable_bonds_sum_fullpolymerfeaturizer_absolute_difference_signed_difference_product_squared_distance_euclidean_distance_max: 64.0
  molecular_weight_sum_fullpolymerfeaturizer_absolute_difference_signed_difference_product_squared_distance_euclidean_distance_min: 174.097940208
  num_hbond_donors_sum_fullpolymerfeaturizer_absolute_difference_signed_difference_product_squared_distance_euclidean_distance_min: 1.0
  num_hbond_acceptors_sum_fullpolymerfeaturizer_absolute_difference_signed_difference_product_squared_distance_euclidean_distance_min: 0.0
  num_rotatable_bonds_sum_fullpolymerfeaturizer_absolute_difference_signed_difference_product_squared_distance_euclidean_distance_min: 8.0
  molecular_weight_sum_fullpolymerfeaturizer_absolute_difference_signed_difference_product_squared_distance_euclidean_distance_sum: 94592.58793238651
  num_hbond_donors_sum_fullpolymerfeaturizer_absolute_difference_signed_difference_product_squared_distance_euclidean_distance_sum: 6.0
  num_hbond_acceptors_sum_fullpolymerfeaturizer_absolute_difference_signed_difference_product_squared_distance_euclidean_distance_sum: 9.0
  num_rotatable_bonds_sum_fullpolymerfeaturizer_absolute_difference_signed_difference_product_squared_distance_euclidean_distance_sum: 108.0

Adding Terminal Groups to Polymers¶

In [8]:

polymer = Polymer.from_psmiles("[*]=C(C#N)NC(=O)c1ccc(C(=O)NC(=[*])C#N)cc1")
# Add terminal groups for the sidechain and backbone
polymer.backbone_terminal_groups = {"[*]": "*O"} 
polymer.sidechain_terminal_groups = {"[*]": "*CCO"}

polymer = Polymer.from_psmiles("[*]=C(C#N)NC(=O)c1ccc(C(=O)NC(=[*])C#N)cc1") # Add terminal groups for the sidechain and backbone polymer.backbone_terminal_groups = {"[*]": "*O"} polymer.sidechain_terminal_groups = {"[*]": "*CCO"}

In [9]:

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

backbone_multi_featurizer = MultipleFeaturizer(backbone_featurizers)
features = backbone_multi_featurizer.featurize(polymer)
labels = backbone_multi_featurizer.feature_labels()

# Print labels and features
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) features = backbone_multi_featurizer.featurize(polymer) labels = backbone_multi_featurizer.feature_labels() # Print labels and features for label, value in zip(labels, features): print(f"{label}: {value:.2f}")

numbackbonefeaturizer: 1.00
num_rings_sum_with_terminalgroups_backbonefeaturizer: 1.00
num_atoms_sum_with_terminalgroups_backbonefeaturizer: 16.00
topological_surface_area_sum_with_terminalgroups_backbonefeaturizer: 92.34

In [10]:

sidechain_featurizers = [
    NumSideChainFeaturizer(),
    SideChainFeaturizer(NumAtoms()),
    SideChainFeaturizer(NumHBondDonors()),
    SideChainFeaturizer(TopologicalSurfaceArea()),
]

sidechain_multi_featurizer = MultipleFeaturizer(sidechain_featurizers)
features = sidechain_multi_featurizer.featurize(polymer)
labels = sidechain_multi_featurizer.feature_labels()

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

sidechain_featurizers = [ NumSideChainFeaturizer(), SideChainFeaturizer(NumAtoms()), SideChainFeaturizer(NumHBondDonors()), SideChainFeaturizer(TopologicalSurfaceArea()), ] sidechain_multi_featurizer = MultipleFeaturizer(sidechain_featurizers) features = sidechain_multi_featurizer.featurize(polymer) labels = sidechain_multi_featurizer.feature_labels() # Print labels and features for label, value in zip(labels, features): print(f"{label}: {value:.2f}")

numsidechainfeaturizer: 2.00
num_atoms_with_terminalgroups_sidechainfeaturizer_sum: 10.00
num_hbond_donors_with_terminalgroups_sidechainfeaturizer_sum: 2.00
topological_surface_area_with_terminalgroups_sidechainfeaturizer_sum: 88.04

In [11]:

# Full polymer featurizers
full_polymer_featurizers = [
    FullPolymerFeaturizer(NumAtoms()),
    FullPolymerFeaturizer(NumHBondDonors()),
    FullPolymerFeaturizer(TopologicalSurfaceArea()),
]
full_multi_featurizer = MultipleFeaturizer(full_polymer_featurizers)
features = full_multi_featurizer.featurize(polymer)
labels = full_multi_featurizer.feature_labels()
print("\nFull molecular features:")
for label, value in zip(labels, features):
    print(f"{label}: {value:.2f}")

# Full polymer featurizers full_polymer_featurizers = [ FullPolymerFeaturizer(NumAtoms()), FullPolymerFeaturizer(NumHBondDonors()), FullPolymerFeaturizer(TopologicalSurfaceArea()), ] full_multi_featurizer = MultipleFeaturizer(full_polymer_featurizers) features = full_multi_featurizer.featurize(polymer) labels = full_multi_featurizer.feature_labels() print("\nFull molecular features:") for label, value in zip(labels, features): print(f"{label}: {value:.2f}")

Full molecular features:
num_atoms_sum_with_terminalgroups_fullpolymerfeaturizer: 20.00
num_hbond_donors_sum_with_terminalgroups_fullpolymerfeaturizer: 2.00
topological_surface_area_sum_with_terminalgroups_fullpolymerfeaturizer: 105.78