ChEMBL

Practical information

Content scope

• The page describes ChEMBL as a database of bioactive drug-like small molecules containing 2-D structures, calculated properties such as logP and molecular weight, and abstracted bioactivities.
  chembl.gitbook.io · 2026-11-13

  ChEMBL is a database of bioactive drug-like small molecules, it contains 2-D structures, calculated properties (e.g. logP, Molecular Weight, Lipinski Parameters, etc.) and abstracted bioactivities (e.g. binding constants, pharmacology and ADMET data).

  https://chembl.gitbook.io/chembl-interface-documentation/about · Dataset page · checked 2026-11-13

• The text states the data is abstracted and curated from the primary scientific literature and covers a significant fraction of the SAR and discovery of modern drugs.
  chembl.gitbook.io · 2026-11-13

  The data is abstracted and curated from the primary scientific literature, and cover a significant fraction of the SAR and discovery of modern drugs

  https://chembl.gitbook.io/chembl-interface-documentation/about · Dataset page · checked 2026-11-13

Identifiers

• ChEMBL_IDThe text states that the Substructure and Similarity services accept SMILES, InChI Key and molecule ChEMBL_ID as arguments.
  chembl.gitbook.io · 2026-11-13

  These methods accept SMILES, InChI Key and molecule ChEMBL_ID as arguments and in the case of similarity searches an additional identity cut-off is needed.

  https://chembl.gitbook.io/chembl-interface-documentation/web-services/chembl-data-web-services · Dataset page · checked 2026-11-13

• → other resources (unspecified)The X-Ref endpoint is described as providing cross references to other resources for compounds.
  chembl.gitbook.io · 2026-11-13

  Cross references to other resources for compounds

  https://chembl.gitbook.io/chembl-interface-documentation/web-services/chembl-data-web-services · Dataset page · checked 2026-11-13

Access

• Bulk downloadftp.ebi.ac.uk SDF · FASTA · MySQL · PostgreSQL · SQLite · FP · H5 · XML · TXT · PNG
  • Own terms: commercial use Not stated · redistribution With conditions
    ftp.ebi.ac.uk · 2026-11-13

    The data content in ChEMBL is licensed under a highly permissive Creative Commons license - specifically the "CC Attribution-ShareAlike 3.0 Unported license", see LICENSE file. The required attribution should contain the url of the ChEMBL resource, and also the release version, e.g.:

    https://ftp.ebi.ac.uk/pub/databases/chembl/ChEMBLdb/latest/README · README · checked 2026-11-13

  • ChEMBL flags questionable activity values in ACTIVITIES.data_validity_comment (e.g. 'Outside typical range', 'Non standard unit for type', 'Potential transcription error', 'Potential author error'); ChEMBL leaves it to users to decide whether to keep them.
    chembl.gitbook.io · 2026-11-13

    The use of the flag will allow users to decide whether or not to retain those values in their results set after running a search on the interface or using the downloaded database.

    https://chembl.gitbook.io/chembl-interface-documentation/frequently-asked-questions/chembl-data-questions · Dataset page · checked 2026-11-13

• Web APIebi.ac.uk XML · JSON · YAML · SVG · SDF

  Same terms as the card's license.

  • The text states that GET parameters must be encoded into the URL and that URL length is limited, so POST requests are often more convenient.
    chembl.gitbook.io · 2026-11-13

    In GET request all the parameters has to be encoded into URL. Because there is a limitation of how long a URL can be it's often more convenient to use POST requests instead.

    https://chembl.gitbook.io/chembl-interface-documentation/web-services/chembl-data-web-services · Dataset page · checked 2026-11-13

• Bulk downloadchembl.gitbook.io turtle

  Same terms as the card's license.

• Web APIchembl.gitbook.io

  Same terms as the card's license.

• Database dumpftp.ebi.ac.uk

  Same terms as the card's license.

• Web interface onlychembl.gitbook.io

  Same terms as the card's license.

Versions

• LatestThe text states the version of ChEMBL is chembl_37.
  ftp.ebi.ac.uk · 2026-11-13

  ChEMBL data is from http://www.ebi.ac.uk/chembl - the version of ChEMBL is chembl_37.

  https://ftp.ebi.ac.uk/pub/databases/chembl/ChEMBLdb/latest/README · README · checked 2026-11-13

• Breaking changeSchema changes between releases: ChEMBL 37 deprecated tables CURATION_LOOKUP and PROTEIN_CLASS_SYNONYMS and fields CELL_DICTIONARY.CL_LINCS_ID and ASSAYS.CURATED_BY, and added ACTIVITIES.MODALITY.
  ftp.ebi.ac.uk · 2026-11-13

  # Tables that have been deprecated since ChEMBL 36: CURATION_LOOKUP PROTEIN_CLASS_SYNONYMS

  https://ftp.ebi.ac.uk/pub/databases/chembl/ChEMBLdb/latest/chembl_37_release_notes.txt · official-release-notes · checked 2026-11-13

• Breaking changeApproved-drug status (MAX_PHASE) changed markedly in ChEMBL 36 after a stricter reading of ATC and BNF sources; 169 compounds were restored to MAX_PHASE 4 in ChEMBL 37, and ChEMBL says differences between recent releases will persist.
  ftp.ebi.ac.uk · 2026-11-13

  Differences in ATC MAX_PHASE will still exist between recent ChEMBL releases, as this curation effort is ongoing.

  https://ftp.ebi.ac.uk/pub/databases/chembl/ChEMBLdb/latest/chembl_37_release_notes.txt · official-release-notes · checked 2026-11-13

Known defects and misuse

• ChEMBL flags likely repeat citations of an earlier measurement in the potential_duplicate column.
  chembl.gitbook.io · 2026-11-13

  Identification of potential duplicate values - where an activity measurement is likely to be a repeat citation of an earlier measurement, rather than an independent measurement (flagged using the potential_duplicate column).

  https://chembl.gitbook.io/chembl-interface-documentation/frequently-asked-questions/chembl-data-questions · Dataset page · checked 2026-11-13

• Landrum and Riniker: combining ChEMBL IC50 values from different assays with minimal curation gives poor agreement (about 65% of pairs differ by >0.3 log units, 27% by >1 log unit); stricter metadata matching reduces but does not remove the noise, and Ki shows similar noise.
  pmc.ncbi.nlm.nih.gov · 2026-11-13

  IC 50 assays selected using minimal curation settings have poor agreement with each other: almost 65% of the points differ by more than 0.3 log units, 27% differ by more than one log unit

  https://pmc.ncbi.nlm.nih.gov/articles/PMC10934815/ · peer-reviewed-paper · checked 2026-11-13

• Kalliokoski et al. reached a milder conclusion: the standard deviation of ChEMBL IC50 data is only about 25% larger than for Ki, so mixing IC50 from different assays adds a moderate amount of noise.
  pmc.ncbi.nlm.nih.gov · 2026-11-13

  The standard deviation of IC 50 data is only 25% larger than the standard deviation of K i data, suggesting that mixing IC 50 data from different assays, even not knowing assay conditions details, only adds a moderate amount of noise to the overall data.

  https://pmc.ncbi.nlm.nih.gov/articles/PMC3628986/ · peer-reviewed-paper · checked 2026-11-13

• Kramer et al. found ChEMBL Ki data contain unit-transcription errors, undifferentiated stereoisomers and repeated citations (90% of all pairs), and estimated an experimental-uncertainty ceiling of R² about 0.81.
  pubmed.ncbi.nlm.nih.gov · 2026-11-13

  Careful filtering of the data was required because ChEMBL contains unit-transcription errors, undifferentiated stereoisomers, and repeated citations of single measurements (90% of all pairs).

  https://pubmed.ncbi.nlm.nih.gov/22643060/ · peer-reviewed-paper · checked 2026-11-13

• ChEMBL 37 removed 13 duplicate documents found within and across the BindingDB and SureChEMBL patent sources.
  ftp.ebi.ac.uk · 2026-11-13

  13 intra-source and inter-source duplicate documents were identified.

  https://ftp.ebi.ac.uk/pub/databases/chembl/ChEMBLdb/latest/chembl_37_release_notes.txt · official-release-notes · checked 2026-11-13

Suited for

• The text states that the Substructure and Similarity web service resources allow the chemical content of ChEMBL to be searched.
  chembl.gitbook.io · 2026-11-13

  The 'Substructure' and 'Similarity' web service resources allow for the chemical content of ChEMBL to be searched.

  https://chembl.gitbook.io/chembl-interface-documentation/web-services/chembl-data-web-services · Dataset page · checked 2026-11-13

• Landrum and Riniker caution that combining literature IC50 data from different assays into one ML set is scientifically risky, more so for large literature collections like ChEMBL.
  pmc.ncbi.nlm.nih.gov · 2026-11-13

  the risks of assays being incompatible are even higher when pulling data from large collections of literature data like ChEMBL.

  https://pmc.ncbi.nlm.nih.gov/articles/PMC10934815/ · peer-reviewed-paper · checked 2026-11-13

Not suited for

• The text states that the Image resource, unlike the other resources, does not accept filtering and paging arguments.
  chembl.gitbook.io · 2026-11-13

  Unlike the other resources it does not except filtering and paging arguments, but does accept image specific arguments.

  https://chembl.gitbook.io/chembl-interface-documentation/web-services/chembl-data-web-services · Dataset page · checked 2026-11-13