Mark J. Daly, Ph.D., is an Associate Professor at the Massachusetts General Hospital/Harvard Medical School and an Associate Member of the Broad Institute. This lecture explains how statistical and computational methods can aid researchers in understanding, diagnosing, and treating disease. Association mapping is the process identifying genetic variation which can explain phenotypic variation, which is particularly important for understanding disease phenotypes (e.g., susceptibility). Historically, the method of choice for solving this problem was linkage analysis. However, advances in genomic technology have allowed for a more powerful method called genome-wide association.
More recent advances in technology and genomic data have allowed for novel integrative analyses which can make powerful predictions about diseases. Any discussion about the basis of disease must consider both genetic and environmental effects. However, it is known that many traits, for example those in Figure 30.1, have significant genetic components. Formally, the heritability of a phenotype is the proportion of variation in that phenotype which can be explained by genetic variation. The traits in Figure 30.1 are all at least 50% heritable. Accurately estimating heritability involves statistical analyses on samples with highly varied levels of shared genetic variation (e.g., twins, siblings, relatives, and unrelated). Studies on the heritability of Type 2 diabetes, for example, have shown that given you have diabetes, the risk to the person sitting next to you (an unrelated person) increases by 5–10%; the risk to a sibling increases by 30%; and the risk to an identical twin increases by 85%–90%.