Enabling a data-driven approach to drug discovery
We collaborate with the world’s leading pharma and biotech companies to lower the risks and time frames in drug discovery by enabling a data-driven approach to target discovery, indication selection and target evaluation.
We provide ready-to-use, AI-driven applications and capabilities for discovering, as well as, evaluating potential novel gene-disease combinations that:
Predicting novel gene-disease associations using computational disease models
Euretos has developed a data-driven, rational approach towards target discovery and indication selection using computational disease models. These are machine learning models that are trained on the known genetic associations for diseases across the whole indication landscape. They leverage biological gene networks inferred from co-expression and protein-protein interactions to calculate a gene-disease score for each protein-coding gene by:
Based on this approach, the gene-disease score quantifies a gene’s association with the disease, even when there is no direct evidence.
The gene-disease prediction scores can be used to rank all genes as a potential target for a disease, providing users with a rational and systematic approach towards target discovery and indication selection. A p-value is estimated based on a comparison to randomly permuted gene networks.
In the indication ranking results, the separate scores for the biological networks of the target are provided to help assess the contribution of each type of molecular interaction to the gene-disease association. This approach provides data-driven insights on novel indications while minimising literature bias.
Empowering domain experts to draw their own conclusions
The ability to assess the predicted target-indication associations by domain experts is crucial to move a candidate forward in the selection process.
We enable this expert evaluation in two important ways.
First of all we break down the single gene-disease score per type of gene network. For each gene, its interactions in various types of biological molecular networks are evaluated such as co-expression correlation, protein – protein interactions, pathways associations. For each of these networks we provide the individual scores so domain experts get a sense of which type of molecular interactions provide the strongest contribution to the gene-disease association.
Secondly, we provide domain experts access to the Euretos AI Platform which provides an AI-integrated knowledge base that harmonises supporting evidence from biomolecular databases, publications and patents.
The Euretos platform gives access to key data such as biological interaction networks, gene expression in tissues, cell types and tumors, pathway interactions, regulatory and signaling responses, clinical variant associations with disease and phenotypes, clinical trial outcomes, target tractability and patents.
This approach empowers researchers to further assess the outcomes of the predictive computation disease models and draw their own, evidence-based conclusions.