Bulk data

The only way of downloading up-to-date SureChEMBL data.

Bulk data are machine-readable and allow any user to download the SureChEMBL data.

Content

The bulk data offer access to the whole collection of SureChEMBL compounds with the patents in which they have been extracted from.

compounds

All the compounds in SureChEMBL

COLUMN_NAME	DATA_TYPE	COMMENT
id	INT64	Compound unique identifier (PK)
smiles	STRING	Canonical smiles, generated using RDKit
inchi	STRING	IUPAC standard InChI for the compound
inchi_key	STRING	IUPAC standard InChI key for the compound
mol_weight	DOUBLE	Molecular weight of the full compound including any salts

patents

All the patents containing compounds in SureChEMBL.

COLUMN_NAME	DATA_TYPE	COMMENT
id	INT64	Patent unique identifier (PK)
patent_number	STRING	Standardised format used to search system. Format: CC-PATNO-KK, e.g. WO-2011161255-A2
country	STRING	Publication country
publication_date	DATE	The date when the patent was first published
family_id	INT64	Data retrieved from the EPO. Generally, simple families contain all records which share the same priority. -1 if we have not received a family ID from DOCDB.
cpc	LIST OF STRINGS	Cooperative Patent Classification (full list)
ipcr	LIST OF STRINGS	International Patent Classification Reform (full list)
ipc	LIST OF STRINGS	International Patent Classification (full list)
ecla	LIST OF STRINGS	European Classification (replaced by CPC in 2013)
assignee	LIST OF STRINGS	Assignee refers to the person or legal entity who owns the entire right, title, and interest in the application. Note that this field contains the assignee information provided by the publishing authority and in most cases doesn’t reflect reassignments.
title	STRING	Title of the document

patent_compound_map

Patent-compound relationships with the patent field where the compound is found.

COLUMN_NAME	DATA_TYPE	COMMENT
patent_id	INT64	Patent unique identifier (FK)
compound_id	INT64	Compound unique identifier (FK)
field_id	INT64	Field unique identifier (FK)

fields

patent field descriptions

COLUMN_NAME	DATA_TYPE	COMMENT
id	INT64	Field unique identifier (PK)
fieldname	INT64	Patent field where information can be found 1 - Description 2 - Claims 3 - Abstract 4 - Title 5 - Image (for patents after 2007) 6 - MOL attachments (US patents after 2007)

biomedical_entities

All the non chemical annotation in SureChEMBL. We provide the captured terms, the corrected terms and the resolved form when the entity can be mapped to an ontology or controlled vocabulary.

COLUMN_NAME	DATA_TYPE	COMMENT
id	INT64	Biomedical entity unique identifer (PK)
type_id	INT64	Unique biomedical entity type identifier (FK)
original_text	STRING	Text entity as it was captured in the patent
corrected_text	STRING	Corrected text
resolved_form	STRING	Identifier from a relevant data source based on the entity type. 1 - Uniprot & HGNC 2 - MeSH, Disease Ontology & Wikipedia

biomedical_locations

COLUMN_NAME	DATA_TYPE	COMMENT
id	INT64	Biomedical entity location unique identifier (PK)
entity_id	INT64	Biomedical entity unique identifer (FK)
patent_id	INT64	Patent unique identifier (FK)
field_id	INT64	Field unique identifier (FK)
count	INT64	Entity-Document-Field triplate count

biomedical_types

COLUMN_NAME	DATA_TYPE	COMMENT
id	INT64	Unique biomedical entity type identifier (PK)
type_name	STRING	Leadmine entity type name
description	STRING	Entity type description

Format

We produce the datasets in Parquet format that allows us to expose nested information in a machine-readable way. Parquet is a columnar storage format widely used in big data platforms like Apache Spark and Hadoop. It’s designed for efficient querying, particularly for analytics that access only a subset of columns. Benefits include faster read performance, better compression, and support for complex nested data.

While interacting with Parquet files requires using a library (e.g., Pandas, PyArrow, Polars, DuckDB in Python), it’s very similar to querying a SQL database.

Note: we no longer provide an out-of-the-box solution for database creation (SureChEMBL data client no longer gets data update).

Access

The parquet files are available here.

Frequency of release

The bulk data are updated every 2 weeks with new parquet files containing the whole dataset. Because every release is independent from the previous one, the data schema might change to offer more data. In such case, the users will be notified.

Usage

Examples (available soon) of how to use the bulk data are available in this notebook. Ultimately, the data can be loaded in a database.