The search for genetic variants that influence medically-relevant phenotypes has been driven by the central dogma of genetics: DNA encodes genes as discrete units of inheritance; genes encoded by DNA are transcribed to RNA; messenger RNA molecules are translated to proteins; one protein has one function; and aberrant proteins encoded by mutant gene variants lead to aberrant biological activity. This paradigm explains many human diseases [1] and has been extended to diseases where two or perhaps three genes contribute [2]. Unfortunately, the paradigm of ‘one gene, one protein, one function’ has failed to reveal the causes of prevalent complex diseases known conclusively from twin studies to have a genetic basis [3,4]. These diseases include arthritis, diabetes and risk factors for cardiovascular disease. Similarly, this paradigm may also fail for traits related to pharmacokinetics and pharmacodynamics. Technological advances in genetic analysis now permit whole-genome association, in which the entire genome of an individual is screened for genetic variation at a series of preselected locations, termed markers. In these studies, there are approximately as many markers as individual genes, if not more. SNPs are the default choice for markers, and SNP sets for the human genome now have 100,000–500,000 markers to provide coverage of our 24,000 genes. Testing these markers for association with a phenotype is a challenging signal-to-noise problem: it is difficult to distinguish between the few real causative effects and the statistical flukes that are expected from examining 100,000 markers, whether markers are individual SNPs or correlated local patterns of SNPs termed haplotypes [5]. It is now possible to test hypotheses that phenotypes result from cumulative, subtle regulatory changes, in contrast to the catastrophic failure of a single protein as in classic Mendelian genetics (Figure 1). Rather than linking a phenotype to a single gene, the new paradigm may be linking a trait to a pathway or a biological system. Although linking genetics to pathways is an intriguing idea, the lack of systematic pathway maps beyond basic metabolism has presented an insurmountable obstacle. Consequently, most work in human genetics turned to other avenues; the theoretical framework for pathway-level mapping is an important current avenue of research. Only recently, in part owing to the author’s work [6], have pathway maps become available to the extent that they can be used in genetic analysis.