We suspect that many biologists who have been following the development of functional 'omics and systems biology are wondering why, compared with the success in integrating information from large-scale studies of transcripts and, more recently, proteins, there is less success in integrating information about metabolism.Information about thousands of transcripts has been integrated into large databases that are used by thousands of scientists (Zimmermann et al., 2004). Rapid advances will soon allow a similar approach in quantitative proteomics. Strategies also exist to integrate information about the expression of genes, as transcripts, with information about changes in the levels of the encoded proteins. Even when changes in transcript levels do not lead to changes in the levels of the encoded proteins, this apparent discrepancy can often be explained by quantitative analysis of the data sets and information about translational regulation and protein turnover (Piques et al., 2009; Schwanhäusser et al., 2011). Such advances are providing systemslevel information about transcriptional and signaling networks that regulate the transcript and protein abundance. However, another important goal of systems biology is a comprehensive understanding of the functional importance of these changes in transcript and protein abundance. Although there is sometimes an encouragingly simple relationship between changes of transcripts and proteins and changes in downstream biological functions, quite often there is a major discordance. In this scientific correspondence, we will discuss reasons for this apparent disconnect, which lie partly in technical issues and partly in the complexity and structure of metabolic networks, and then propose some priorities in metabolic research to combat these problems.As we climb up what has been described as the "pyramid of life" (Barabási and Oltvai, 2004) from gene through RNA and protein to phenotype (which includes chemical composition), we are confronted with an increasing complexity both at the chemical level and in the logic that is needed to draw inferences from one level to the next. Nucleic acids consist of four bases, joined by the same phosphoester bond and with pairwise interacting side chains. This simple structure allows a near-to-binary storage and transmission of information. Proteins are chemically much more complex, due to the functionalities of 20 different amino acid side chains. This large expansion in chemical complexity lies at the center of the flexibility and complexity of biological systems, as it allows different proteins to perform an enormous range of different functions. Nevertheless, proteins still have a simple primary structure, with amino acids linked via a peptide bond in a fixed sequence that, via the universal genetic code, can be unambiguously linked to the sequence of nucleic acids (Aebersold and Mann, 2003;Fernie et al., 2004; Baerenfaller et al., 2008). This logic breaks down when we consider entities that are generated by the activities of individual pr...