In the biological sciences, the phenotype represents the traits that are observable in an individual. As there are potentially many such traits, we speak of multivariate phenotype to describe a large-dimensional ensemble of related traits that are of interest in a specific study.

Examples of multivariate phenotypes include brain imaging data, comprehensive datasets of anthropometric traits (sex, age, weight, etc), or traits that relate to a disease syndrome (blood pressure, lung function indices, etc).

In order to make sense of multivariate phenotypes, we must correlate them with explanatory variables such as genetic markers. This requires finding the linear combination of phenotypes with the lowest noise or error under some constraint. Hence an estimate of the covariance matrix of phenotype errors is required. When the number of phenotypes is large, the proprietary techniques of Studdridge International can help estimate this covariance matrix more accurately, and hence extract more useful information from the available data.