The vast majority of human diseases do not result from single genetic changes. Most diseases are the consequence of multiple molecular alterations, genetic and otherwise, with an environmental component frequently also playing a role. What is more, even 'single-gene' diseases are likely to be modulated by alleles at other loci as well as by nongenetic effects. Systems biology—the use of comprehensive large-scale data to understand biology at a more global scale—is therefore widely seen as necessary for a complete understanding of disease.

Disease as a systems-level phenomenon.

There are many different ways in which systems approaches are being used for this purpose. In the study of cancer, for instance, several projects have been initiated worldwide to map the genomic, transcriptomic and epigenomic changes that occur in human cancer, and these have already yielded maps of glioblastoma, breast and pancreatic cancer genomes, among others. Challenges abound, however, owing in part to the difficulty of distinguishing between 'driver' and 'passenger' events, and to the notorious heterogeneity of cancer, both within a given tumor and between individuals with the disease.

Separately, it has been shown that gene expression profiling data can be combined with information from the human protein-protein interaction network to yield more predictive markers of breast cancer prognosis than is possible with lists of differentially expressed genes alone. It should, however, be recognized that these network-based markers are still far from perfect.

Yet another instance in which systems information may prove useful for understanding the molecular basis of disease is in the more precise identification of causal disease genes, or groups of genes, from candidates suggested by genome-wide association studies. Existing approaches are largely based on bioinformatics analyses, and these are likely to develop further, but experimental strategies using high-throughput imaging to test candidate genes are a potentially exciting complement for the future, in cases for which useful cellular assays can be designed.

The tremendous interest in the systems biology of disease notwithstanding, both the use of networks as reliable biomarkers of disease and that of integrated 'omics' approaches to profile and describe diseased cells and tissues are still in their early stages. Watch for methods development and much movement in this area in the coming years!