henhaobio
Sign in

Nucleic Acids ResearchBioinformaticsPeer-reviewed journal article

Integration of the Drug-Gene Interaction Database (DGIdb) with open crowdsource efforts

Sharon L. FreshouriD, Susanna KiwalaiD, Kelsy C. CottoiD, Adam C. CoffmaniD, Joshua F. McMichaeliD, Jonathan J SongiD, Malachi GriffithiD, Obi L. GriffithiD, Alex Handler WagneriD

PDFDOIPreprintCodeData
In-depth full-text digest available · Jump to digest ↓

Abstract

The Drug-Gene Interaction Database (DGIdb, www.dgidb.org) is a web resource that provides information on drug-gene interactions and druggable genes from various sources including publications, databases, and other web-based sources in one resource. These drug, gene, and interaction claims are normalized and grouped to identify aliases, merge concepts, and reduce redundancy. The information contained in this resource is available to users through a straightforward search interface, an application programming interface (API), and TSV data downloads. DGIdb 4.0 is the latest major update of this database. Seven new sources have been added, bringing the total number of sources included to 41. Of the previously aggregated sources, 15 have been updated. DGIdb 4.0 also includes improvements to the process of drug normalization and grouping of imported sources. Other notable updates include further development of automatic jobs for routine data updates, more sophisticated query scores for interaction search results, extensive manual curation of interaction source link outs, and the inclusion of interaction directionality. A major focus of this update was integration with crowd-sourced efforts, including leveraging the curation activities of Drug Target Commons, using Wikidata to facilitate term normalization, and integrating into NDEx for producing network representations.

In-depth summary

Claude Opus 5.5 · xhigh effort · 8 Nov 2026 · from the full text of version 1 (20 Sept 2020) · CC BY 4.0

DGIdb 4.0 aggregates drug-gene interaction and druggability claims from 41 sources, adding crowd-sourced Drug Target Commons, Wikidata-based drug normalization, search-relative query scores, and NDEx export.

Drug-gene interaction and druggability evidence is spread across dozens of databases with redundant drug aliases; in DGIdb 3.0, static downloads lagged monthly source updates and pairwise rule-based drug grouping required full re-grouping whenever sources changed. Version 4.0 adds monthly releases, a per-source drug normalizer built on Wikidata and the Variant Interpretation for Cancer Consortium thera-py package, and specificity-aware scores.

Key results

  1. ShownDGIdb now draws on 41 sources. It added crowd-curated Drug Target Commons and COSMIC as interaction sources, added five druggable-category sources, and refreshed 15 existing sources.

    New records: 23,916 drug-gene interaction claims and 8,463 druggable gene category claims. New category sources: Tempus xT, IDG top-priority genes, Human Protein Atlas, Oncomine, Pharos. Updated sources include DrugBank, Guide to Pharmacology, Gene Ontology, OncoKB, PharmGKB and Therapeutic Target Database. · Abstract; NEW AND UPDATED CONTENT; Figure legends

  2. ShownInteraction searches now return a query score relative to the search set. It boosts partners shared by several queried entities and down-weights partners with many interactions. The static interaction score now also accounts for how many partners each drug and gene has.

    The query score combines publication and source evidence, the number of queried genes (or drugs) that hit the same partner, and the average partner count divided by that partner's own count. MEK1-Cobimetinib scored 9.76 with MEK1/MEK2 queried and rose when BRAF and KRAS were added. ZEB1-Doxorubicin (87 Doxorubicin interactions) scores far below ZEB1-Salinomycin, Salinomycin's only interaction. · UPDATED QUERY AND INTERACTION SCORES; Figure legends

  3. ClaimedDrug grouping no longer uses pairwise rules applied to the whole database. It now runs source by source through a reusable normalizer built on VICC thera-py, to which the team contributed a Wikidata normalizer, so community edits can improve how drug names are matched.

    Drug claims were normalized with the thera-py ChEMBL and Wikidata normalizers. Precedence rules favor primary labels before aliases, exact-case before case-insensitive matches, and ChEMBL before other normalizers. Pseudocode for building merged concepts is in Supplementary Data. No accuracy metrics are reported. · IMPROVEMENT OF DRUG NORMALIZATION APPROACH; SUMMARY AND FUTURE DIRECTIONS

  4. ShownRoutine maintenance is automated. More sources run as scheduled online updaters under Sidekiq, TSV downloads are released monthly with historic snapshots kept, and the DGIdb network is uploaded to NDEx every month.

    12 of 41 sources use online updaters (PharmGKB was converted and Pharos added as one), including Entrez (genes 41,102 to 42,767) and Ensembl (version 90_38 to 101_38). Sidekiq replaced DelayedJob, with Airbrake error alerts and GitHub CI testing. · NEW AND UPDATED CONTENT; MONTHLY DATA RELEASES; INTEGRATION WITH NDEX AND OTHER EXTERNAL…; TECHNOLOGY IMPROVEMENTS AND IMPLEMENTATION

  5. ShownEvery source's license was manually curated and listed on the sources page, with specially licensed sources flagged on the downloads page. Where known for an interaction type, results now show inhibitory or activating directionality, plus link-outs to specific interaction evidence when available.

    Licensing changes since 3.0 are recorded: PharmGKB moved to CC BY-SA 4.0, DrugBank to a custom non-commercial license, OncoKB API access to approved non-commercial research only, and JAX-CKB API access to negotiated licenses only. Restricted data are exported separately. Example: chaperone interactions are flagged as activating. · IMPROVED TRANSPARANCY AND DETAILS ON LICENSING…; OTHER SEARCH RESULT IMPROVEMENTS

Methods, data and reproducibility

DGIdb 4.0 imports drug, gene, interaction and druggable-category claims from 41 sources through TSV importers and scheduled online importers. It normalizes drugs per source with thera-py ChEMBL and Wikidata normalizers, groups genes on Entrez concepts, and serves results through a web interface, an API, monthly TSV releases and NDEx.

  • DGIdb 4.0 aggregated drug, gene, interaction and druggable gene category claims · n = 41 sources (seven new, 15 updated)
  • New interaction sources: Drug Target Commons (crowd-sourced expert curation) and COSMIC (cancer mutations and drug resistance) · n = 23,916 new drug-gene interaction records
  • New druggable category sources: Tempus xT panel, IDG top-priority genes, Human Protein Atlas, Oncomine assay, Pharos · n = 8,463 new druggable gene category records
  • Updated existing sources including DrugBank, Guide to Pharmacology, Gene Ontology, OncoKB, PharmGKB, Therapeutic Target Database · n = 15 updated sources
  • Entrez gene concepts, the core reference for gene grouping · n = 42,767 genes (up from 41,102)
  • Ensembl gene aliases, updated from version 90_38 to 101_38
  • ChEMBL and Wikidata drug records used through thera-py normalizers for drug concept grouping
Methods & tools
DGIdb 4.0, thera-py (VICC therapy normalization package), thera-py ChEMBL normalizer, thera-py Wikidata normalizer, Sidekiq job scheduler, NDEx (Network Data Exchange), Ruby on Rails 6, PostgreSQL 12, Airbrake exception tracking, GitHub continuous integration, Drug Target Commons
Software & environment
Rails 6, Ruby 2.6.5, PostgreSQL 12 on an Ubuntu 20.04 LTS server; Sidekiq job scheduling; GitHub CI testing against multiple Ruby and PostgreSQL versions; source code under the MIT license.
Reproducibility gaps
Some sources have restricted licenses (OncoKB API limited to approved non-commercial research, JAX-CKB to negotiated licenses), so their data are exported separately or users are sent to the original source. Normalization pseudocode is only in Supplementary Data.
Limitations (6)
  • The public API is still separate from the internal code that renders the web pages. This leaves redundant code and a less complete API; the authors plan to merge the two. · stated by the authors
  • Online updaters cover 12 of 41 sources; the authors plan to convert more where possible and to finish migrating TSV-based sources off the legacy DSL importer style. · stated by the authors
  • License restrictions on sources such as OncoKB and JAX-CKB mean some data cannot be offered as DGIdb downloads; users have to get it from the providers. · stated by the authors
  • Directionality is displayed only when it is defined for an interaction type, and evidence-specific link-outs only where the source provides them, so these details are absent for some results. · stated by the authors
  • The new query and interaction scores are shown only through worked examples (such as MEK1-Cobimetinib 9.76). No benchmark or user study shows that they rank relevant drugs or genes better than the earlier static score. · noted in this summary
  • The redesigned drug normalization is not measured: there is no grouping precision, coverage of normalized drug claims, or comparison with the earlier pairwise grouping. · noted in this summary

Read this if you query DGIdb for drug-gene interactions or druggable genes and need to know what changed in version 4.0: sources, licensing, scoring and drug normalization.

Citations
1,003
FWCI
95.99
Field percentile
top 1%

Reproducibility ledger

Nobody has reported trying to reproduce this paper yet. Did you replicate an experiment or run the code? Your report saves the next lab a week.

Sign in to file a reproducibility report.

Related papers

Similarity Map
SelectedSame fieldCross-field
THISCotto '17Cannon '26Schimmelpfe '26Mondal '25Grissa '22Deng '18
89% semantic matchBioinformatics
Kelsy C. Cotto, Alex Handler Wagner, Yang-Yang Feng+7 more·Nucleic Acids Research·Nov 6, 2017· 888 citations