SIDER Reference Ingest Guide¶
Source Information¶
InfoRes ID: infores:sider
Description: SIDER (Side Effect Resource) is a publicly available database of marketed drugs and their recorded adverse drug reactions (side effects). Side-effect information is extracted from public documents and drug package inserts / product labels, and side-effect terms are mapped to standard medical vocabularies (MedDRA, with UMLS concept identifiers). SIDER groups side effects by drug, and where available also captures the frequency with which a side effect is reported. The most recent release (version 4.1) contains information on 1,430 drugs and 5,868 adverse drug reactions. SIDER is developed and maintained at EMBL.
Citations: - Kuhn M, Letunic I, Jensen LJ, Bork P. The SIDER database of drugs and side effects. Nucleic Acids Research. 2016;44(D1):D1075-9. DOI: 10.1093/nar/gkv1075
Data Access Locations: - SIDER download page - http://sideeffects.embl.de/download/
Data Provision Mechanisms: file_download
Data Formats: tsv
Data Versioning and Releases: SIDER version 4.1 was released in 2015 and is the latest release. Updates are irregular and infrequent, and there has been no new release in many years. Versioning uses a simple major.minor scheme (currently 4.1).
Additional Notes: ['As the latest release (4.1) dates from 2015, SIDER may not include recent drug approvals or complete/updated adverse event profiles. Ongoing maintenance appears limited.']
Ingest Information¶
Ingest Categories: primary_knowledge_provider
Utility: Provides drug to adverse drug reaction (side effect) associations useful for pharmacovigilance, drug safety analysis, and adverse drug reaction prediction studies in Translator use cases.
Scope: Covers marketed drugs and their documented adverse drug reactions as found on drug product labels, as reported by SIDER. This ingest currently takes only the label-specific side effects file and only its MedDRA preferred term (PT) rows.
Relevant Files¶
| File Name | Location | Description |
|---|---|---|
| meddra_all_label_se.tsv.gz | http://sideeffects.embl.de/media/download/meddra_all_label_se.tsv.gz | Side effects specifically found on drug labels, with STITCH compound identifiers, MedDRA concept type (LLT/PT), UMLS concept identifiers, and side-effect names. |
Included Content¶
| File Name | Included Records | Fields Used |
|---|---|---|
| meddra_all_label_se.tsv.gz | Label-specific side-effect associations whose MedDRA_concept_type is PT (preferred term); duplicate subject-predicate-object triples are collapsed to a single edge. | STITCH_compound_id_stereo, MedDRA_concept_type, UMLS_concept_id, side_effect_name |
Filtered Content¶
| File Name | Filtered Records | Rationale |
|---|---|---|
| meddra_all_label_se.tsv.gz | Records whose MedDRA_concept_type is LLT (lower level term) are excluded; only PT (preferred term) rows are retained. | Per the SIDER README (http://sideeffects.embl.de/media/download/README): there is at least one PT for every LLT, but sometimes the PT is the same as the LLT. LLTs are sometimes too detailed, so PT rows are retained to avoid overly granular, redundant side-effect terms. |
Future Content Considerations¶
edge_property_content: Some background exploration suggested that SIDER provides qualitative frequency information about side effects (e.g. terms like 'very common', 'uncommon'), available in the meddra_freq.tsv file. This seems like useful information we should consider capturing as an edge property.
node_property_content: Chemical (drug) nodes are currently emitted with an identifier only and no name. SIDER's drug_names.tsv file maps STITCH/PubChem compound identifiers to common drug names and could be used to populate biolink:name on the chemical nodes. - Relevant files: drug_names.tsv
Target Information¶
Edge Types¶
| Subject Categories | Predicate | Object Categories | Knowledge Level | Agent Type | UI Explanation |
|---|---|---|---|---|---|
| biolink:ChemicalEntity | biolink:DiseaseOrPhenotypicFeature | knowledge_assertion | manual_agent | SIDER reports side-effect information extracted from drug product labels, regulatory documents, and official package inserts, mapping side-effect terms to standard vocabularies such as MedDRA (with UMLS concept identifiers). The SIDER record used here reports that the drug [SUBJECT] can cause the side effect [OBJECT]. This is represented in Translator using the 'has_side_effect' predicate. |
Node Types¶
| Node Category | Source Identifier Types | Additional Notes |
|---|---|---|
| biolink:ChemicalEntity | PUBCHEM.COMPOUND | Chemical identifiers are derived from SIDER STITCH compound identifiers (CID...), which are stripped to their numeric PubChem Compound id and prefixed with PUBCHEM.COMPOUND. No name is currently set on chemical nodes (see future considerations). |
| biolink:DiseaseOrPhenotypicFeature | UMLS |
Future Modeling Considerations¶
edge_properties: Consider capturing SIDER's qualitative side-effect frequency information (e.g. 'very common', 'uncommon') as an edge property, if it is added to the ingest (see the ingest content future considerations).
Provenance Information¶
Contributors: - Vlado Dancik - code author, domain expertise - Kevin Schaper - code support - Evan Morris - code support - Sierra Moxon - data modeling, code support - Matthew Brush - data modeling, domain expertise