Methodological issues and experimental design considerations in metabolic profile-based classifications

Abstract

The onset of the era of -omics and systems biology arguably brings with it fundamental shifts in every aspect of biological research. At the conceptual level, there is a shift in focus from mechanistic details to more generalistic hypotheses that is followed by data-driven (discovery-based) set of analyses. A similar shift from univariate statistical analyses to multivariate approaches to data analysis such as those which have been more commonly used in physical sciences and engineering is also apparent. In reality, however, successful implementation of these new approaches relies as much on experimental design and mechanistic understanding, as ever. Similarly, mathematical analysis eventually returns to an understanding of whether or not observed phenomena are primarily and inherently a result of statistical sampling or biological reality. © 2005 Springer Science+Business Media, Inc. All rights reserved.