Finding new targets for drug discovery

The Open Targets consortium is a public-private partnership, consisting of GSK, Biogen, Takeda, Celgene, Sanofi, the Wellcome Sanger Institute and the European Bioinformatics Institute (EMBL-EBI). The goal is systematic identification, validation and prioritization of possible drug targets by combining data from a variety of sources.

 

Open Targets Platform

Target-disease associations have proven to be very useful in drug discovery and drug repurposing. The Open Targets Platform is a web application that provides an interface to the data that is being aggregated and reprocessed on a regular basis from an increasing list of valuable public data sources. It allows the user to query a target to find all diseases associated with it, or a disease to find all targets associated with it.

(Click on photo to visit the page)

The data is presented in an intuitive and straightforward overview. Associations are scored based on the strength of the evidence, derived from multiple publicly available data sources. All the associations are integrated and sorted accordingly in the overview.

The user can drill down to individual evidence items that support the association. Evidence items have a link to their original source where more detailed information can be found.

 

Diseases are all mapped to the Experimental Factor Ontology (EFO) and can be associated to targets on multiple levels. You can for example find all targets for type II diabetes mellitus or find all targets for diabetes mellitus, that will include targets for all the subtypes (see image below).

 

Open Targets Genetics

The Genetics Portal is a companion to the target validation platform and allows you to explore variant to gene and disease associations, obtained from genome-wide association studies (GWAS). It has a search box similar to the one on the Open Targets Platform front page, which allows you to search by gene, variant or trait.

 

There are several dashboards to start exploring the data, depending on what you search for. For instance, when searching for a gene you will find a list of variants that were found to be correlated with a disease or trait in one of the studies, ordered by p-value. Not that this list does not only include variants that are on the gene but also variants that are found to be related to the gene, for instance because of linkage disequilibrium or through the fine-mapping, based on methods like eQTL or pQTL.

 

When querying a variant you will see related genes (e.g. through eQTL/pQTL), associated traits (e.g. from UK Biobank), a list of studies, a list of lead variants and a list of tag variants.

 

When searching for a trait/disease you will see a genome wide overview of independently-associated loci.

 

What are the benefits

  • Great platform for proprietary data

These tools allow users to quickly and easily look into all data that is available on a certain gene or disease. They are supported by a strong consortium and there is a wealth of public data available, that can be combined with your proprietary data.

  • Data centralization

There are many, many different strategies to identify and validate targets, all with their specific advantages and disadvantages. There is no way to harmonize data from all these strategies into one platform without either losing important details or creating a complicated and expensive piece of software that nobody will be able to work with.

With the Open Targets Platform the results of your experiments will be harmonized. These could come from experiments that are completely different in their materials and methods. This way the harmonization effort is small while the platform remains easy to use. Besides, you still have the benefits of having a catalog of everything that is going on in target discovery across your organisation. This will allow users to dig into the details by following the links to the actual experimental data.

  • Fit for ML/AI

Thanks to the data curation and harmonization that takes place before data enters the platform and a great REST API, the Open Targets Platform is very well suited for machine learning purposes. Interestingly, if you have algorithms that predict new targets, the output of these algorithms can also be loaded into the platform. This will allow users to share these results across the organisation and present them in the context of already available data.

 

What we do

The Hyve supports clients in the installation, customization and maintenance of the Open Targets Platform and Genetics Portal and the loading of proprietary data. We can extend the platform to support data types that are specific to your organisation.

 

Contact us for any further information.