Events and Meetings of Italian Statistical Society, Advances in Latent Variables - Methods, Models and Applications

Font Size: 
A general framework for modelling covariances
Age Klaas Smilde, Marieke E Timmerman, Edoardo Saccenti, Jeroen J Jansen, Huub C.J. Hoefsloot

Last modified: 2013-06-16


In many cases in systems biology data sets are available from observing the same system under different conditions. The classical way to analyse such data sets is by using univariate or multivariate methods and look for differences. These analyses are then performed on the data sets directly. Sometimes the information in the data is in terms of changes of covariances where such a covariance is representative for one condition. Then it is not sufficient anymore to perform data analysis directly on the data using univariate or multivariate tools, but (changes in) covariances have to be modelled. We recently developed a new method – coined COMSCA – especially suited for functional genomics data. We will present a general framework for these covariance modeling methods showing examples of some of these methods illustrated with real-life metabolomics data.

Full Text: PDF