Single nucleotide polymorphisms (SNPs) are the most common form of genetic variation between human individuals. The majority of monogenic disease is mediated by this mechanism, and it is believed that susceptibility to polygenic diseases and individual response to medication can also be understood largely in these terms. Large scale SNP mapping is rapidly increasing the amount of data available on SNPs within the human population. To understand these data and exploit them for development of new therapies requires models of the link between SNPs and function. We propose to develop such a model for SNPs that act via effects on protein function. Most monogenic disease SNPs operate in this manner so it is expected that the results will have wide applicability for interpreting and utilizing human SNP information. The model is based on the large amount of data available on the effect of single residue mutations in vitro. Those data can be understood in terms of the situation of the mutation within the protein structure. The mutation data, together with the structural context, have been used to devise a set of rules that identify which coding region SNPs are potentially harmful in vivo. We have tested a first version of the model against a set of data on SNPs known to cause human disease and a sample of SNPs from the human population. The results provide insights into the mechanism of action of SNPs. We now propose to develop a set of software capable of performing a full analysis of all disease related and general population SNP data. The software will be used to perform an analysis of the extent of harmful SNPs in the human population, and to identify a subset of these likely to be involved in polygenetic diseases.
