drug-resolver 0.0.5

A tool to resolve messy drug names in biomedical datasets to standard identifiers

Drug Resolver

This package offers a simple way to resolve drug names to their corresponding PubChem entries.

This is useful for identifying unknown drug synonyms or matching drug synonyms across different datasets in an automated way.

Installation

pip install drug-resolver

Usage

from drug_resolver import resolve_drug

d1 = resolve_drug("tetracycline hcl")
print(d1)
"""
ResolvedDrug(
   name='Tetracycline hydrochloride', 
   pubchem_compound_id='54704426', 
   pubchem_substance_id='103591391', 
   smiles='CN(C)C1C(=O)C(C(N)=O)=C(O)C2(O)C(=O)C3=C(O)c4c(O)cccc4C(C)(O)C3CC12.Cl', 
   drug_classes=None, 
   fda_approval=<FDAApprovalStatus.APPROVED: 'APPROVED'>, 
   pubchem_parent_compound_id='54675776')
"""

d2 = resolve_drug("Sumycin")
print(d2)
"""
ResolvedDrug(
    name='Tetracycline', 
    pubchem_compound_id='54675776', 
    pubchem_substance_id='124766046', 
    smiles='CN(C)C1C(=O)C(C(N)=O)=C(O)C2(O)C(=O)C3=C(O)c4c(O)cccc4C(C)(O)C3CC12', 
    drug_classes=frozenset({'Established Pharmacologic Class [EPC] - Tetracycline-class Antimicrobial'}), 
    fda_approval=<FDAApprovalStatus.APPROVED: 'APPROVED'>, 
    pubchem_parent_compound_id=None)
"""

assert d1 == d2 # True

The equality operator between ResolvedDrug objects will return True under the following conditions:

• The pubchem_compound_id attribute is the same
• The pubchem_substance_id attribute is the same
• The smiles strings refer to the same molecule
• Two compounds share the same pubchem_parent_compound_id or the pubchem_compound_id is the same as the pubchem_parent_compound_id of the other compound.

Caching

You can set the env variable DRUG_RESOLVER_REQUEST_CACHE to a filename. A sqlite database will be created at the filename and all PubChem requests will be cached there. This is useful for avoiding repeated requests to the PubChem API. By default, the cache will be stored in memory and will be lost when the program exits.

Here is an example of the cache works:

(.venv) drug-name-resolver % DRUG_RESOLVER_REQUEST_CACHE=/tmp/test.sqlite3 ipython
Python 3.11.11 (main, Dec  3 2024, 17:20:40) [Clang 16.0.0 (clang-1600.0.26.4)]

In [1]: from drug_resolver import resolve_drug

In [2]: %time d1 = resolve_drug("tetracycline hcl")
CPU times: user 48.1 ms, sys: 10.7 ms, total: 58.8 ms
Wall time: 1.53 s
(.venv)  drug-name-resolver % ls -lh /tmp/test.sqlite3
-rw-r--r--@ 1 quinnj2  wheel   468K Jan 30 14:51 /tmp/test.sqlite3
(.venv) drug-name-resolver % DRUG_RESOLVER_REQUEST_CACHE=/tmp/test.sqlite3 ipython
Python 3.11.11 (main, Dec  3 2024, 17:20:40) [Clang 16.0.0 (clang-1600.0.26.4)]

In [1]: from drug_resolver import resolve_drug

In [2]: %time d1 = resolve_drug("tetracycline hcl")
CPU times: user 24.2 ms, sys: 4.06 ms, total: 28.3 ms
Wall time: 29.3 ms