J. Claiborne Stephens

Deconstructing the relationship between genetics and race

The success of many strategies for finding genetic variants that underlie complex traits depends on how genetic variation is distributed among human populations. This realization has intensified the investigation of genetic differences among groups, which are often defined by commonly used racial labels. Some scientists argue that race is an adequate proxy of ancestry, whereas others claim that race belies how genetic variation is apportioned. Resolving this controversy depends on understanding the complicated relationship between race, ancestry and the demographic history of humans. Recent discoveries are helping us to deconstruct this relationship, and provide better guidance to scientists and policy makers.

The predictive power of haplotypes in clinical response

A variety of approaches have been proposed to find genetic markers that can be used in a clinical setting. Single nucleotide polymorphisms (SNPs) are the basis of the most commonly used approaches. Here we describe an approach using gene-based haplotypes, which are collections of SNPs located throughout the ftinctional regions of candidate genes, and organised as they occur separately on an individuals two chromosomes. The main point of this review is that the haplotype has greater power than any individual SNP to track an unobsenrved, but evolutionarily linked, variable site.

SNP and haplotype variation in the human genome.

We have surveyed and summarized several aspects of DNA variability among humans. The variation described is the result of mutation followed by a combination of drift, migration and selection bringing the frequencies high enough to be observed. This paper describes what we have learned about how DNA variability differs among genes and populations. We sequenced functional regions of a set of 3950 genes. DNA was sampled from 82 unrelated humans: 20 African-Americans, 20 East Asians, 21 Caucasians, 18 Hispanic-Latinos and 3 Native Americans. Different aspects of variability showed a great deal of concordance. In particular, we studied patterns of single nucleotide polymorphism (SNP) allele and haplotype sharing among the four, large sample populations. We also examined how linkage disequilibrium (LD) between SNPs relates to physical distance in the different populations. It is clear from our findings that while many variants are common to all populations, many others have a more restricted distribution. Research that attempts to find genetic variants that explain phenotypic variants must be careful in their choice of study population.

How many SNPs does a genome-wide haplotype map require?

We derive and compare several estimates of the number of SNPs that would be required to form the basis of a complete haplotype survey of the human genome. Our estimates make use of reports published by Stephens et al. [1], Patil et al. [2] and Daly et al. [3]. The estimated number of SNPs required for a genome-wide haplotype survey ranges from 180K (based on a European sample of 16 chromosomes) to 600K (based on an ethnically diverse sample of 164 chromosomes). We discuss the implications of using cohorts of different size and ethnic composition and the usefulness of public SNP databases for this effort. Finally, we estimate the experimental effort and cost required to complete a genome-wide haplotype survey.

Notes from the SNP vs. haplotype front

Single nucleotide polymorphisms (SNPs) and haplotypes are commonly used genetic markers in clinical studies. We provide some broad guidelines for deciding which of the two is most appropriate in particular circumstances. Molecular haplotyping techniques are also briefly reviewed and contrasted with electronic approaches.

The structure of common genetic variation in United States populations.

The common-variant/common-disease model predicts that most risk alleles underlying complex health-related traits are common and, therefore, old and found in multiple populations, rather than being rare or population specific. Accordingly, there is widespread interest in assessing the population structure of common alleles. However, such assessments have been confounded by analysis of data sets with bias toward ascertainment of common alleles (e.g., HapMap and Perlegen) or in which a relatively small number of genes and/or populations were sampled. The aim of this study was to examine the structure of common variation ascertained in major U.S. populations, by resequencing the exons and flanking regions of 3,873 genes in 154 chromosomes from European, Latino/Hispanic, Asian, and African Americans generated by the Genaissance Resequencing Project. The frequency distributions of private and common single-nucleotide polymorphisms (SNPs) were measured, and the extent to which common SNPs were shared across populations was analyzed using several different estimators of population structure. Most SNPs that were common in one population were present in multiple populations, but SNPs common in one population were frequently not common in other populations. Moreover, SNPs that were common in two or more populations often differed significantly in frequency from one population to another, particularly in comparisons of African Americans versus other U.S. populations. These findings indicate that, even if the bulk of alleles underlying complex health-related traits are common SNPs, geographic ancestry might well be an important predictor of whether a person carries a risk allele.

DNA variability of human genes

We have investigated the level of DNA-based variation (both SNPs and haplotypes) for several thousand human genes. In addition, we have characterized how this variation is distributed in a number of biologically and clinically important ways. First, we have determined how SNPs are distributed within human genes: where they occur relative to various functional regions; levels of variability of human SNPs; pattern of the molecular sequence of SNPs; and how these compare with the corresponding sequence of a chimpanzee. Second, we have determined how these aspects of SNP distribution vary among four human population samples. All genes were sequenced on DNA obtained from 82 unrelated individuals: 20 African-Americans, 20 East Asians, 21 European-Americans, 18 Hispanic-Latinos and three Native Americans. In particular, we looked at patterns of SNP and haplotype sharing among the four larger population samples. Third, we have determined the patterns of linkage disequilibrium among SNPs, which also determines the haplotype variability of each gene. These characteristics also vary substantially among populations. A deeper understanding of these aspects of human genetic variation will be of vital importance when trying to identify the genetic contribution to complex phenotypes such as aging.

Gene and protein domain-specific patterns of genetic variability within the G-protein coupled receptor superfamily.

AbstractIntroduction: Guanine nucleotide binding proteins (G-proteins) represent the targets for >50% of all therapeutics. There is substantial interindividual variation in response to agonists and antagonists directed to these receptors, which may, in part, be due to genetic polymorphisms. As a class, the sequence variability of G-protein-coupled eceptor (GPCR) genes has not been characterized. Study design: This variability was investigated by sequencing promoter, 5′- and 3′-UTR, coding blocks, and intron-exon boundaries, of 64 GPCR genes in an ethnically diverse group of 82 individuals. Results: Of the 675 single-nucleotide variations found, 61% occurred in ≥1% of the population sample and the nature of these 412 single nucleotide polymorphisms (SNPs) was assessed. 5′-UTR (p = 0.002) and coding (p = 0.006) SNPs were observed more often in GPCR genes, compared with 309 non-GPCR genes similarly interrogated. The prevalence of non-synonymous coding SNPs was unexpectedly high, with 65% of GPCR genes having at least one. Intron-containing genes had half as many non-synonymous coding SNPs compared with intronless genes (p = 0.0009), suggesting that when introns are not available coding regions provide sites for variation. A distinct relationship between the prevalence of non-synonymous SNPs and receptor structural domains was evident (p = 0.0006 by ANOVA), with variability being most prominent in the transmembrane spanning domains (38%) and the intracellular loops (24%). Phosphoregulatory domains, particularly the carboxy terminus, often the site for agonist-promoted phosphorylation by G-protein coupled receptor kinases, were the least polymorphic (8%). Conclusions: There is substantial genetic variability in potentially pharmacologically relevant coding and noncoding regions of GPCRs. Such variability should be considered in the development of new agents, or optimization of existing agents, targeted to these receptors.

Assessing Human Variation Data for Signatures of Natural Selection

In this article, we highlight some of the different types of natural selection, their effects on patterns of DNA variation, and some of the statistical tests that are commonly used to detect such effects. We also explain some of the relative strengths and weaknesses of different strategies that can be used to detect signatures of natural selection at individual loci. These strategies are illustrated by their application to empirical data from gene variants that are often associated with differences in disease susceptibility. We briefly outline some of the methods proposed to scan the genome for evidence of selection. Finally, we discuss some of the problems associated with identifying signatures of selection and with making inferences about the nature of the selective process.

HUMAN GENOME VARIATION: DISEASE, DRUG RESPONSE, AND CLINICAL PHENOTYPES

With the completion of a rough draft of the human genome sequence in sight, researchers are shifting to leverage this new information in the elucidation of the genetic basis of disease susceptibility and drug response. Massive genotyping and gene expression profiling studies are being planned and carried out by both academic/public institutions and industry. Researchers from different disciplines are all interested in the mining of the data coming from those studies; human geneticists, population geneticists, molecular biologists, computational biologists and even clinical practitioners. These communities have different immediate goals, but at the end of the day what is sought is analogous: the connection between variation in a group of genes or in their expression and observed phenotypes. There is an imminent need to link information across the huge data sets these groups are producing independently. However, there are tremendous challenges in the integration of polymorphism and gene expression databases and their clinical phenotypic annotation This is the third session devoted to the computational challenges of human genome variation studies held at the Pacific Symposium on Biocomputing 1,2 . The focus of the session has been the presentation and discussion of new research that promises to facilitate the elucidation of the connections between genotypes and phenotypes using the data generated by high -throughput technologies. Nine accepted manuscripts comprise this year’s original work presented at the conference.