Open Targets Platform 25.06 has been released!

Release Notes Jun 18, 2025

The latest release of the Platform — 25.06 — is now available at platform.opentargets.org.

We’re trying something new: tomorrow at 3pm UK time, join our walkthrough and Q&A session on LinkedIn. The team will showcase the new features and data in this release, and then will answer your questions about the release and the Platform. See you there!

Key points

The latest data from the GWAS Catalog brings in 36% more GWAS credible sets, most of which are from the VA Million Veteran Program study
You can now view association evidence for a target’s interactors directly in our disease associations pages
The Platform data downloads page is now powered by the Croissant metadata standard and easier to navigate
We’ve updated the molecular structure viewer on target profile pages and added one to variant profile pages with AlphaMissense pathogenicity scores
We have incorporated gene burden data from the Broad CVDI Human Disease Portal
Our pharmacogenetics widgets contain additional annotation on the effect of the variants on drug availability
Other improvements: search filters and EFO measurement terms

For the full list of updates, take a look at the release notes. For a list of key stats and metrics, see the Open Targets Community.

36% more credible sets from GWAS Catalog data

Thanks to the merge of Open Targets Genetics data and analysis pipelines into the Open Targets Platform, we can easily pull, finemap, and analyse GWAS data. Where possible, we fine-map and colocalise publication results curated and available through the GWAS Catalog.

The latest GWAS Catalog data in this release adds 36% more credible sets to the Platform from 211 publications.

One of these publications is a large multi-ancestry GWAS meta-analysis for osteoarthritis, combining 87 datasets. Published in Nature in April this year, Hatzikotoulas, Southam, Stefansdottir, Boer, McDonald, et al. integrated their GWAS findings with functional genomics data from relevant tissues, identifying 700 genes with high confidence of being involved in osteoarthritis, and eight biological processes key to the disease development, including circadian clock and glial cell functions.

GWAS credible sets from Hatzikotoulas et al. (GCST90566795). Our analyses established 203 95% GWAS credible sets associated with this study.

VA Million Veteran Program data

80% of new credible sets in this release derive from studies from the VA Million Veteran Program, an ongoing prospective cohort study and the largest multipopulation biobank to date. Verma, Huffman, Rodriguez, Conery, Liu et al. (2024) conducted Genome Wide Association Studies with over 635,000 participants, identifying more than 26,000 variant-trait associations across 1270 traits.

Our processing of the results from this publication contributed 172,000 credible sets that inform our Locus-to-Gene predictions. This included the first genetic evidence in the Platform for the association between ESR1 and osteoporosis.

GWAS associations widget for ESR1 on the osteoporosis associations page. ESR1 is a known target for osteoporosis, with approved drugs such as Estradiol. GWAS data from the VA Million Veteran Program is the first genetic evidence available in the Platform that supports this association. Both GWAS and eQTL colocalisations support the link between ESR1 and osteoporosis.

Changes to GWAS Catalog data in the Platform

We have made the decision to remove case-case studies from our GWAS data, for a total of 77 studies. Case-case studies are difficult to map to the correct disease, and therefore confound target-disease associations.

Nominating potential targets using target interactors

This release comes with a new Target Interactors view, which allows you to access association data for interactors of a given target, to quickly pinpoint other — potentially more favourable — targets, within the same network.

Target interactors can be viewed directly within the main ‘Associations on the Fly’ page. This is an option available for each target on a disease associations page; note that the interaction data is disease-agnostic.

Users can choose one of four sources of molecular interactors — IntAct (binary physical interactions), Reactome (pathway-based interactions), Signor (directional, causal interactions), and String (functional interactions) — and view the association evidence for the top scoring molecular interactors for that database. We’ve set default cutoff interactions scores which you can adjust.

Target interactors subview for IFNAR1 in systemic lupus erythematosus (SLE). The subview shows target-disease associations for targets identified by IntAct as interacting with IFNAR1, and you can view evidence for the association of each interactor. Here, the view is displaying GWAS associations evidence linking TYK2 to SLE. The Platform does not have direct genetic evidence to link IFNAR1 to SLE, but there is an approved therapy, Anifrolumab, targeting IFNAR1 for SLE. Browsing interactors data shows that IFNAR1 interacts with TYK2, which has a well-known genetic association to SLE.

Find out more about target interactors in our documentation.

New data downloads page and Croissant

We have revamped our data downloads page, making it easier for you to find the dataset you need. In addition to the description, each of the 37 datasets now features a data category tag, to make it easier for you to filter. You can view the schema for each dataset, including column descriptions, primary and foreign keys, and the data access options.

The data downloads page for Open Targets Platform datasets. The datasets are now easier to navigate and filter, with detailed schemas and column descriptions.

Croissant

We have described our data following the Croissant metadata standard format, based on JSON-LD, developed by ML-Commons. The Croissant format simplifies how data is used by ML models, and will make Open Targets Platform data discoverable and FAIRer.

Molecular structure viewers

We have updated the Molecular Structure viewer on target profile pages, and added a version of the viewer on missense variant pages, which indicates the location of the variant in the AlphaFold model.

On target pages, the Molecular Structure viewer allows you to browse structural protein information from Uniprot, and compare structures derived from different methods, including AlphaFold predictions, X-ray, EM and NMR.

Molecular Structure widget for TYK2, displaying the AlphaFold predicted structure. You can click through the list to see structures of TYK2 determined by X-ray methods.

For predicted missense variants, we have included a Molecular Structure viewer on the variant page that locates the variant in the AlphaFold structure. Moreover, users have the option to switch to a pathogenicity view, which shows the AlphaMissense pathogenicity for the substitution corresponding to the variant, and the average AlphaMissense pathogenicity score across all possible amino acid substitutions at other positions.

Molecular Structure viewer for 4_1804392_G_A. The confidence view (top) shows the location of the variant in the AlphaFold-predicted protein structure, while the pathogenicity view (bottom), shows the AlphaMissense pathogenicity for the substitution corresponding to the variant, and the average AlphaMissense pathogenicity score for other positions.

Burden evidence from the Broad CVDI Human Disease Portal

We have incorporated data from the Broad CVDI Human Disease Portal into our gene burden analyses. Our Gene Burden datasource incorporates results from the aggregation of rare and ultra-rare variants at the gene-level.

The Portal features pan-ancestry sequencing data from three large biobanks: the UK Biobank, All of Us, and the Mass General Brigham Biobank. Jurgens, Wang, et al. (2024) performed gene-based rare variant testing across almost 750,000 individuals, including more than 155,000 with non-European ancestry, and identified 363 significant associations for 123 genes in 165 diseases, including some novel associations.

In addition to the independent gene burden analyses performed on each cohort, the team also meta-analysed rare variation across ancestries, correcting for overlap in samples between ancestries. This work yielded new associations, but also identified key genes in the phenome of human diseases. These genes are highly pleiotropic and are associated with large effect sizes.

We have mapped the study’s disease terms to our ontology, and processed all burden results from this study that analysed all three cohorts, with different masks and filtered based on the statistical method. This means that certain associations are supported by multiple evidences from this study, if the association was found statistically significant in multiple statistical methods. In total, we have included 1,530 evidence strings for 520 targets/disease pairs in the Platform.

Gene burden widget for YLPM1 and personality disorder in the Open Targets Platform, a novel association brought in by analysis of the Broad CVDI Human Disease Portal. The Portal reports a significant association between rare variants in YLPM1 and bipolar disorder, and personality disorders.

Directionality annotations for pharmacogenetics data

Pharmacogenetics widgets on the variant, target, and drug profile pages now have an additional Directionality column. Where information is available, this provides an indication of how the variant affects drug response.

Pharmacogenetics widget on the VKORC1 profile page. Clicking on the link in the Directionality column opens a drawer with annotation information from PharmGKB through the European Variation Archive.

The European Variation Archive team has been working to integrate more detailed variant-level annotations from PharmGKB and mapping them to clinical annotations. We now have additional information for 67% of all pharmacogenetics evidence in the Platform.

With this additional annotation, you can quickly see whether the variant is associated with a decrease or increase in drug availability, how many studies have information on this relationship, and how concordant they are. There are links to each publication so that you can find out more.

Improved search functionality

We have strengthened our search functionality with performance optimisations and additional filtering capabilities. For example, when you search for a term in the Platform, you now have the option of filtering by entity type.

Measurement EFO terms

The latest Experimental Factor Ontology (EFO) replaced measurement terms with terms from the Ontology of Biological Attributes (OBA), and this is now reflected in the Platform, with downstream consequences on GWAS credible sets and association evidence from Europe PMC.

As usual, please share any comments, questions, or suggestions on the Open Targets Community, and join our walkthrough and Q&A session on LinkedIn!