A blood sample contains a large amount of information about the donor. Full blood counts tap into a small but medically important subset of this information such as the relative proportions of several types of white blood cells in a blood sample: neutrophils, lymphocytes, monocytes, eosinophils, and basophils. Their relative proportions often reflect what is going on the body. For example, high neutrophil counts may indicate a bacterial infection and high eosinophil counts an allergic response. Patterns of methylated and unmethylated cytosines in genomic DNA extracted from white blood cells provide another source of information in a blood sample. These patterns can not only differentiate between cells of different types but also between individuals with different specific phenotypes such as age and BMI and exposure histories such as smoking behaviour. Some phenotypes and exposures have such distinct patterns that the patterns can be used as proxies for the original phenotype or exposure. Proxies for cell counts are the easiest to develop because methylation patterns differ dramatically between cell types. These cell type specific patterns include so much of the genome that they often obscure patterns linked to other phenotypes and exposures. Fortunately in some cases the pattern is only partially obscured and can be rescued by including cell count information in mathematical models. Although directly measured cell counts are unavailable in ALSPAC, they are being generated as part of the ALSPAC Focus @ 24/25 YP Clinic for a subset of the ALSPAC young people. We plan to use these measurements to derive and validate an algorithm for deriving DNA methylation proxies for cell counts in peripheral blood.