BioGRID: Curated Protein & Genetic Interaction Guide

BioGRID (the Biological General Repository for Interaction Datasets) is an open-access biomedical database archiving curated physical, genetic, and chemical interactions, post-translational modifications, and CRISPR phenotype screens. Developed for systems biology and functional genomics, BioGRID indexes experimentally validated macromolecular relationships from primary literature across humans and model organisms, grounding every record in manual literature curation and standardized evidence codes.

What it is

BioGRID is a biomedical interaction repository designed to compile, standardize, and disseminate macromolecular relationships identified in peer-reviewed scientific studies. Established to provide life scientists with a central interaction clearinghouse, the resource indexes two principal classes of experimental data:

• Physical interactions: Direct binding events, multi-protein complexes, and biochemical associations demonstrated through assays including affinity capture-mass spectrometry, co-immunoprecipitation, surface plasmon resonance, and two-hybrid screens.
• Genetic interactions: Functional relationships between pairs of genes where double perturbations yield non-additive phenotypic outcomes, such as synthetic lethality, synthetic growth defects, and epistatic rescue.

The platform also curates chemical-protein interactions, post-translational modification (PTM) sites, and functional screens through its companion repository, BioGRID ORCS (Open Repository for CRISPR Screens). Every curated interaction cross-references primary PubMed citations to verify experimental provenance.

Who runs it and what it covers

BioGRID is managed through an academic collaboration led by Mike Tyers (Institute for Research in Immunology and Cancer, Université de Montréal), Software Engineering Lead Chris Stark (Toronto), and Co-PI Kara Dolinski (Princeton University). Project infrastructure is supported by NIH grants #R01 RR024031 and #OD010949.

The repository curates interactions across humans and key model organisms. As of build 5.0.263 (released in November 2026), the database indexes 3,050,286 raw interactions (2,375,285 non-redundant) extracted from 78,024 publications across 93,602 unique genes:

• Physical interactions (all organisms): 2,160,329 raw records (1,671,037 non-redundant) across 90,810 unique genes from 65,423 publications.
• Genetic interactions (all organisms): 889,957 raw records (730,069 non-redundant) across 26,967 unique genes from 18,420 publications.
• Human interactome (Homo sapiens): 1,505,957 raw physical interactions (1,149,531 non-redundant) across 29,662 unique genes and 41,266 publications.
• Baker’s yeast (Saccharomyces cerevisiae): 287,392 raw physical interactions (198,948 non-redundant) and 494,449 raw genetic interactions across 7,223 unique genes from 11,312 publications.
• Fruit fly (Drosophila melanogaster): 82,457 raw physical interactions (69,555 non-redundant) across 11,133 unique genes from 4,396 publications, aligning with dedicated Drosophila research resources.
• Thale cress (Arabidopsis thaliana): 87,166 raw physical interactions (77,633 non-redundant) across 12,356 unique genes from 2,378 publications.

Curated identifiers are normalized to authoritative registries including NCBI Entrez Gene, Ensembl, UniProt, and model organism databases such as SGD, FlyBase, and MGI.

How to search or use it

BioGRID provides web exploration tools and structured downloads. Searching the repository involves six practical steps using official interface elements:

1. Select search mode: On the BioGRID homepage, choose the appropriate search tab: Gene/Identifier Search (for proteins or genes), Publication Search (by PubMed ID or full-text keywords), or Chemical Search (for small-molecule associations).
2. Set organism scope: By default, queries search All Organisms. To avoid cross-species identifier collisions between orthologs, select your target species (such as Homo sapiens or Saccharomyces cerevisiae) from the organism dropdown.
3. Enter query terms and syntax: Enter official gene symbols, systematic names, or accession identifiers. The engine supports wildcard queries (such as STE*). In publication text searches, boolean prefixes apply: + requires a term (AND), - excludes a term (NOT), and double quotes ("") designate exact phrases.
4. Review headers and interaction counts: On the results page, the Annotation Header displays official nomenclature, synonyms, and database cross-links. The overview bar shows counts of physical associations (orange) and genetic associations (green), with bars indicating high-throughput (HTP) versus low-throughput (LTP) proportions.
5. Examine evidence weighting: Interactors are ranked by evidence weighting where physical interactions receive 1.5 and genetic interactions receive 1.0 (filtered records receive 0.01). Expanding speech bubbles reveals experimental evidence codes and citation links. Hyperlinked interactor symbols enable circular navigation through connected subnetworks.
6. Export query datasets or download releases: Click the on-page dataset download icon to retrieve interactions matching your active search. For bulk pipelines, access the downloads section to retrieve full release archives in standard formats: PSI-MI XML 2.5 (.psi25), PSI MITAB 2.5 (.mitab), BioGRID TAB 3.0 (.tab3), and JSON.

Strengths and limits

Understanding BioGRID’s role in experimental workflows requires balancing its curation strengths against inherent literature constraints.

Strengths:

• Experimental provenance: Every interaction links directly to a peer-reviewed publication and controlled experimental evidence code, avoiding uncurated algorithmic inferences in the primary dataset.
• Combined physical and genetic dimensions: Archiving both direct protein binding and functional genetic interactions provides vital context for cellular buffering, synthetic lethality, and functional redundancy.
• Permissive licensing and standardization: BioGRID is distributed under an MIT-style open license for free academic and commercial use, with formats conforming to IMEx and Proteomics Standards Initiative (PSI) guidelines supporting institutional FAIR data principles.
• Screen integration via ORCS: Dedicated curation of whole-genome CRISPR screens bridges physical interaction networks with functional phenotypic screens.

Limits:

• Literature publication bias: Curated databases reflect research attention; well-studied disease genes (such as TP53 and EGFR) possess thousands of documented edges, whereas uncharacterized proteins have sparse coverage.
• No computational network inference: BioGRID curates only reported experimental findings. Researchers seeking predicted associations based on sequence synteny, co-expression, or automated text-mining must consult integrated network platforms.
• Monthly build cadence: Interaction updates are deployed in discrete monthly releases; literature published immediately after a build freeze appears in subsequent builds.

Alternatives

Selecting an interaction or pathway database depends on whether an investigation requires experimentally validated binding, genetic interactions, or computational network predictions.

Database	Primary Scope	Evidence Type	Scoring & Prediction	Access & Formats
BioGRID	Physical, genetic, and chemical interactions across model organisms and human	Curated experimental literature (HTP and LTP)	Evidence weighting (1.5 physical / 1.0 genetic); no predicted edges	Web portal, REST API, PSI-MI XML 2.5, BioGRID TAB 3.0, JSON
STRING database	Functional and physical protein association networks across thousands of organisms	Integrated primary databases, text-mining, co-expression, genomic synteny	Combined confidence score (0.0 to 1.0) with separate evidence channels	Web interface, REST API, Cytoscape plugin, tabular flat files
IntAct (EMBL-EBI)	Molecular interaction database with deep experimental detail (IMEx partner)	Curated physical interactions with binding domains and kinetics	MI-score (confidence metric based on publication count and methods)	Web portal, PSICQUIC service, PSI-MI XML, MITAB 2.7
MINT	Curated physical protein-protein interactions (IMEx partner)	Curated physical binding verified from peer-reviewed literature	MINT-score calibrated to IMEx standards	Web search, PSI-MI MITAB file downloads
Reactome	Curated biological pathways, reaction cascades, and protein complexes	Peer-reviewed molecular pathways cross-referenced with KEGG	Curated pathway topology and reaction directionality	Web browser, REST API, SBML, BioPAX, tabular exports

FAQ

Can you provide a list of protein-protein interaction databases?

Protein-protein interaction databases fall into three distinct functional categories based on curation methodology:

1. Primary experimental repositories: Databases that manually curate experimentally demonstrated physical binding events from peer-reviewed literature. Major repositories include BioGRID, IntAct (EMBL-EBI), MINT (Molecular INTeraction database), and DIP (Database of Interacting Proteins). Every edge corresponds to documented laboratory assays.
2. Integrated and predictive network platforms: Resources that integrate experimental data with computational predictions, gene co-expression, and automated literature text-mining. The leading platform is the STRING database, which assigns confidence scores across direct physical interactions and functional associations.
3. Pathway and complex repositories: Databases mapping interactions into structured biochemical pathways and multi-protein assemblies, including Reactome, KEGG, and the Complex Portal.

Researchers should rely on primary repositories like BioGRID or IntAct when validating specific physical binding partners, and integrated platforms like STRING for broad functional discovery.

What is the difference between BioGRID and STRING?

The primary difference lies in data provenance. BioGRID exclusively curates experimentally validated physical and genetic interactions extracted manually from peer-reviewed literature without computational inference. STRING imports experimental records (including BioGRID data) and combines them with computational predictions, genomic synteny, and automated PubMed text-mining to score functional relationships across more than 12,000 organisms.

What is BioGRID ORCS?

BioGRID ORCS (Open Repository for CRISPR Screens) is a companion database archiving curated whole-genome CRISPR knockout and activation screens from biomedical literature. It standardizes cell lines, guide RNA libraries, and phenotypic hits, enabling researchers to link physical interaction networks with functional genetic dependencies.

What download formats does BioGRID support?

BioGRID supports standard formats defined by the Proteomics Standards Initiative (PSI), including PSI-MI XML 2.5 (.psi25) and PSI MITAB 2.5 (.mitab). It also provides tabular formats—BioGRID TAB 2.0 and TAB 3.0 (.tab3)—as well as JSON records, post-translational modification tables (PTMTAB/PTMREL), and chemical interaction files (CHEMTAB).

Is BioGRID free for commercial use?

Yes. BioGRID is distributed under an MIT-style open license allowing free academic and commercial use. Downstream applications and publications must cite the primary BioGRID database articles and provide attribution in documentation.

Sources

• The BioGRID Interaction Repository: https://thebiogrid.org/ (Checked November 7, 2026).
• BioGRID Documentation Wiki and About Us: https://wiki.thebiogrid.org/doku.php/aboutus (Checked November 7, 2026).
• BioGRID Database Statistics, Build 5.0.263: https://wiki.thebiogrid.org/doku.php/statistics (Checked November 7, 2026).
• BioGRID Search Tips and Results Layout Guide: https://wiki.thebiogrid.org/doku.php/search_tips and https://wiki.thebiogrid.org/doku.php/results (Checked November 7, 2026).
• BioGRID Download Formats and Licensing Terms: https://wiki.thebiogrid.org/doku.php/downloads and https://wiki.thebiogrid.org/doku.php/terms_and_conditions (Checked November 7, 2026).