Descripción del proyecto
There are thousands of common non-communicable diseases (NCDs) that lack safe and effective drugs despite accounting for most health care expenses and years lived with disability. Me and my team will address this unmet clinical need by developing innovative algorithms that generate and integrate evidence from massive scale genomic studies with real-world data based on millions of patients from electronic health records (EHRs) to identify potent drug targets and opportunities for drug repurposing. To achieve this aim, we will 1) harness ethnically diverse biobanks (>500,000 participants) with whole-exome/genome sequencing and EHR linkage powered by deep learning models to gain new insights into the aetiology of neglected NCDs that are needed for rational drug design, 2) create a genetically anchored biomedical knowledge graph that incorporates rich functional genomic data from single-cell studies with drug characteristics to predict promising drug targets using deep graph neural networks, and 3) establish convergence of genetic and real-world evidence of proposed drug targets by emulating clinical trials in multiple large EHR datasets (>50 million patients). Unique access to diverse hospital cohorts and a clinical trial unit at one of the largest European hospitals, the Charité Universitätsmedizin Berlin, will further accelerate clinical translation for selected examples. With GenDrug, we aim to build a community resource to enable and accelerate drug development using ‘big data’ for hundreds to thousands of diseases that currently lack safe and effective treatments.