Programmatic access to DrugBank drug and target data; use when you need to download, parse, and analyze DrugBank XML for properties, interactions, pathways, and pharmacology.
75%
Does it follow best practices?
Run evals on this skill
Adds up to 20 points to the overall score
Passed
No findings from the security scan
Fix and improve this skill with Tessl
tessl review fix ./scientific-skills/Evidence Insight/drugbank-database/SKILL.mddrugbank-downloader (requires DrugBank access).lxml for reliable extraction of nested DrugBank entities.pandas DataFrames for filtering, joining, and export.networkx (e.g., DDI graphs, drug–target bipartite graphs).rdkit for structure-based processing (e.g., SMILES/InChI handling when present).drugbank-downloader (version varies by your environment)lxml>=4.9pandas>=2.0networkx>=3.0rdkit>=2022.09 (optional; required only for structure/chemistry workflows)"""
End-to-end example:
1) Parse a local DrugBank XML file
2) Extract a minimal drug table
3) Extract drug-drug interactions
4) Build a DDI graph
Prerequisites:
- You must obtain DrugBank XML via your DrugBank account/license.
- Place the XML file at ./drugbank.xml (or update the path).
"""
from lxml import etree
import pandas as pd
import networkx as nx
DRUGBANK_XML_PATH = "./drugbank.xml"
NS = {"db": "http://www.drugbank.ca"} # DrugBank XML namespace
# --- Parse XML ---
tree = etree.parse(DRUGBANK_XML_PATH)
root = tree.getroot()
# --- Extract drug records (minimal fields) ---
drugs = []
for drug in root.xpath("//db:drug", namespaces=NS):
drugbank_id = drug.xpath("string(db:drugbank-id[@primary='true'])", namespaces=NS).strip()
name = drug.xpath("string(db:name)", namespaces=NS).strip()
drug_type = drug.get("type", "").strip()
# Optional: first SMILES if present
smiles = drug.xpath(
"string(db:calculated-properties/db:property[db:kind='SMILES']/db:value)",
namespaces=NS,
).strip()
drugs.append(
{
"drugbank_id": drugbank_id,
"name": name,
"type": drug_type,
"smiles": smiles or None,
}
)
drugs_df = pd.DataFrame(drugs).dropna(subset=["drugbank_id"])
print("Drugs:", len(drugs_df))
print(drugs_df.head())
# --- Extract drug-drug interactions ---
interactions = []
for drug in root.xpath("//db:drug", namespaces=NS):
src_id = drug.xpath("string(db:drugbank-id[@primary='true'])", namespaces=NS).strip()
src_name = drug.xpath("string(db:name)", namespaces=NS).strip()
for ddi in drug.xpath("db:drug-interactions/db:drug-interaction", namespaces=NS):
tgt_id = ddi.xpath("string(db:drugbank-id)", namespaces=NS).strip()
tgt_name = ddi.xpath("string(db:name)", namespaces=NS).strip()
description = ddi.xpath("string(db:description)", namespaces=NS).strip()
if src_id and tgt_id:
interactions.append(
{
"source_id": src_id,
"source_name": src_name,
"target_id": tgt_id,
"target_name": tgt_name,
"description": description or None,
}
)
ddi_df = pd.DataFrame(interactions)
print("Interactions:", len(ddi_df))
print(ddi_df.head())
# --- Build a DDI graph ---
G = nx.from_pandas_edgelist(
ddi_df,
source="source_id",
target="target_id",
edge_attr=["description"],
create_using=nx.Graph(),
)
print("DDI graph nodes:", G.number_of_nodes())
print("DDI graph edges:", G.number_of_edges())
# Example analysis: top 10 drugs by interaction degree
top_degree = sorted(G.degree, key=lambda x: x[1], reverse=True)[:10]
top_degree_df = pd.DataFrame(top_degree, columns=["drugbank_id", "degree"]).merge(
drugs_df[["drugbank_id", "name"]],
on="drugbank_id",
how="left",
)
print(top_degree_df)Access & authentication
drugbank-downloader step is responsible for fetching the release artifacts; ensure you comply with DrugBank terms.XML parsing approach
lxml.etree is used for robust XPath-based extraction.http://www.drugbank.ca); XPath queries must include the namespace mapping (e.g., NS = {"db": "http://www.drugbank.ca"}).Core extraction patterns
db:drugbank-id[@primary='true']db:namedb:calculated-properties/db:property[db:kind='SMILES']/db:valuedb:drug-interactions/db:drug-interaction with fields db:drugbank-id, db:name, db:descriptionData modeling
pandas DataFrames for normalized tables (drugs, targets, interactions, pathways).networkx for graph representations:
description as edge attribute).Performance considerations
etree.iterparse) and writing intermediate results to disk.Further references
references/data-access.mdreferences/drug-queries.mdreferences/interactions.mdaipoch/medical-research-skills
686e09d
If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.