• candidate gene studies
• respiratory disease

With the completion of a working draft of the human genome sequence, we are standing at the threshold of the most exciting period in biomedical research. It is now clear that the human genome contains around 32 000 genes,¹ and information is rapidly accumulating on the degree of interindividual variability in the sequences of these genes through collaborations such as the single nucleotide polymorphism (SNP) consortium.² Differences between individuals in their susceptibility to develop common diseases, the severity of their disease, and individual responses to treatment are all potentially predictable on the basis of the interaction of environmental factors with an individual's particular set of genotypes. Over the last few years this has led to a proliferation of association studies examining candidate gene polymorphisms for their potential role in respiratory disease. These studies are easy to perform—needing only a stored DNA sample and access to a relevant clinical data set—but there are many potential pitfalls to performing candidate gene association studies which have resulted in substantial criticism of this kind of study. This editorial provides guidance on some aspects of the design of such studies.

Some of the major issues which need to be considered in the design of candidate gene association studies are listed in table 1 and are briefly discussed below. It is estimated that SNPs occur at a rate of approximately 1 in 300 base pairs in non-coding DNA and 1 in 600 base pairs in coding DNA. Given that most genes are at least 1 kb in length (and many are much greater), it is immediately apparent that, even considering only the coding regions within the genome, …