AbstractMany naturally occurring genetic variants perturb specific aspects of human biology. Therapeutic medicines are also designed to perturb particular aspects of biology, for example by inhibiting a specific protein. Most current drug discovery paradigms first develop the chemical matter to effect the required perturbation and then test the consequences of this in humans. The overwhelming majority of drug trials fail. Human genetics allows for the possibility of learning about perturbation of human biology in humans. Although still difficult, with the right data and data science it can be possible to use naturally occurring genetic variants to understand the consequences of a particular perturbation in silico before having to develop chemical matter or undertake trials in humans. This provides a novel route into target discovery, understanding of relevant cell types, biological mechanisms, and potentially the identification of the most appropriate patient sub-populations. The talk will illustrate the opportunities and challenges of this kind of approach through a number of examples.