The emergence of new large-scale technologies with increased evidence granularity (e.g. scRNAseq or CRISPR), as well as the increasing interest to stratify diseases based on these or other phenotypic readouts, introduces challenges that the Platform will seek to address in the near future through adaptation of our data model and new ways to represent data. Determining the relative importance of different pieces of evidence when combining the available information to suggest potentially successful targets for drug discovery is a further challenge. The inclusion of new systematic data sources, such as the state-of-the-art GWAS data from Open Targets Genetics Portal, revealed the need to appropriately weight and benchmark scored evidence against other orthogonal data sources, a challenging task due to the lack of appropriate gold standards. Moreover, recent studies have pinpointed to the usefulness of expanding experimentally determined evidence using protein interaction networks (39,40). Network data can help to circumvent issues such as non-tractable targets or safety liabilities, as well as identify functionally linked novel targets with no prior evidence. We are therefore exploring different approaches to further exploit the Platform target–disease evidence in the context of their molecular interactions.
