 There are about 80 different autoimmune diseases and collectively they affect about 5% of the population. And an autoimmune disease arises when the immune system becomes dysregulated and causes inflammation in normal host tissue. Now we have lots of treatments which are available for autoimmune diseases but no treatment works on all patients. We need to be able to better match patients to treatment. This is what we call personalised medicine. And in some cases we need to be able to match patients to treatments that work on individual molecules. This is called precision medicine. So what sort of information and how much information do we require to make a diagnosis? At the moment diagnosis is usually made when a patient consults their physician and the physician performs various tests, blood tests, scans and X-rays. But the diagnosis is approximate and many of the treatments that we use work in many different ways. So we overcome all of this uncertainty, this imprecision by matching patients with these fuzzy diagnoses with treatments that work in multiple ways and we evaluate all of that in a clinical trial. Now autoimmune diseases do have a significant genetic component. We know this because if one identical twin is affected there's a 50% chance that the other identical twin will be affected. And even in families where there are no twins there is still clustering of autoimmune and inflammatory diseases. Now we're all made up of trillions of different cells, different cell types clustered together to form organs. The proteins in the cells are encoded by genes which are arranged on chromosomes and our genetic makeup comprises complementary strands of 3 billion bases A, C, T and G. And if we sequence all 3 billion bases we observe a remarkable person-to-person variation. Some of this variation is of recent onset and is rare and some of the genetic variation has been present in humans and has been propagated through many generations. So is it the rare or the common variants that contribute to autoimmune disease? If we classify patients according in the usual sort of fuzzy way and then look at common genetic variants we find that some of those common genetic variants are even more common in patients with autoimmune disease than they are in the general population. But this hasn't really led to any insights into new treatments and if we add up all of those common variants then they don't really account for the overall genetic liability. So one possible explanation for this misinheritability is that there's a contribution from rare genetic variants as well. The fuzzy diagnostic categories might be fuzzy because they're in fact made up of many subcategories of disease but overall characterizing the complete genetic makeup of an autoimmune disease might not be necessary for understanding or for personalizing medicine and this is because not all genetic variants are pathological. Now we can engineer genetic variants that we identify in humans to alter the orthologous variant in mice. This way we can test whether there is a phenotype conferred by the genetic variant and if there is we can perform experiments to try and understand the mechanism and this is the strategy that we take at the Centre for Personalized Immunology where the discovery program begins with characterizing variants in patients. We then proceed to understand mechanism and by building these bespoke models we can arrive at a decision about a particular genetic variant even if it's only present in one family. But this level of granularity might not be necessary for personalization of medicine it may be that genetic variants converge on pathways of cell signalling and cell differentiation and if this is the case then understanding disease pathways might be important. So rare variants give us insights into pathways of disease but the discovery process is complex and time consuming but if we can elucidate these pathways and if there is convergence of genetic variants then this might reveal the subcategories of disease which are the targets for treatment personalization.