Drug Knowledge Graph

Table of Contents

• Drug Knowledge Graph

  • Drug Transformer

  • Drug Graph

Drug Transformer

PyTrial also offers utilities for processing and transforming drug related data. The first one is pytrial.model_utils.drug.DrugTransformer that works for mapping between different drug terminologies, e.g., from ATC to NDC, from ATC to SMILES, etc.

A colab example is available at demo_drug_utils.ipynb.

from pytrial.model_utils.drug import DrugTransformer

# Create a transformer
dt = DrugTransformer()

# drug name to SMILES
print(dt.name2smiles('Acetylsalicylic acid'))

# NDC to ATC4
print(dt.ndc2atc('00002140701'))

# ATC4 to NDC
print(dt.atc2ndc('C01BA'))

# drug name to NDC
print(dt.name2ndc('NEO*IV*Gentamicin'))

# NDC to drug name
print(dt.ndc2name('63323017302'))

# ATC4 to drug name
print(dt.atc2name('S01HA'))

# drug name to ATC4
print(dt.name2atc('atomoxetine'))

# NDC to SMILES
print(dt.ndc2smiles(['00002140701','00088222033']))

# ATC4 to SMILES
print(dt.atc2smiles(['S01HA','N06AX']))

Drug Graph

pytrial.model_utils.drug.DrugGraph is a class for creating a drug knowledge graph.

A colab example is available at demo_drug_graph.ipynb.

from pytrial.model_utils.drug import DrugGraph

dg = DrugGraph()

print(dg.graph)
'''
DiGraph with 251947 nodes and 303921 edges
'''

The yielded dg.Graph is a networkx.DiGraph object, which can be used to analyze the drug knowledge graph or create a graph neural network model.