Pharmacogenomic variation in the Malagasy population: implications for the antimalarial drug primaquine metabolism.

Aim: Antimalarial primaquine (PQ) eliminates liver hypnozoites of Plasmodium vivax. CYP2D6 gene variation contributes to PQ therapeutic failure. Additional gene variation may contribute to PQ efficacy. Information on pharmacogenomic variation in Madagascar, with vivax malaria and a unique population admixture, is scanty. Methods: The authors performed genome-wide genotyping of 55 Malagasy samples and analyzed data with a focus on a set of 28 pharmacogenes most relevant to PQ. Results: Mainly, the study identified 110 coding or splicing variants, including those that, based on previous studies in other populations, may be implicated in PQ response and copy number variation, specifically in chromosomal regions that contain pharmacogenes. Conclusion: With this pilot information, larger genome-wide association analyses with PQ metabolism and response are substantially more feasible.

Patterns of risk for diabetic retinopathy in the Mumbai slums: The Aditya Jyot Diabetic Retinopathy in Urban Mumbai Slums Study (AJ-DRUMSS) Report 3.

Diabetes onset precedes diabetic retinopathy (DR) by 5-10 years, but many people with diabetes remain free of this microvascular complication. Our aim was to identify risk factors for DR progression in a unique and diverse population, the slums of Mumbai. We performed a nested case-control study of 1163 diabetics over 40 years of age from slums in 18 wards of Mumbai. Data was collected on 33 variables and assessed for association with DR using both univariate and multivariate analyses. Stratified analyses were also performed on males and females, separately. Among hypertensive individuals we also assessed whether duration of hypertension associated with DR. Of 31 non-correlated variables analysed as risk factors for DR, 15 showed evidence of significant association. The most prominent included sex, where being a female associated with decreased odds of DR, while longer duration of diabetes and poor glycaemic control associated with increased odds. The duration of diabetes effect was partially, but significantly, mediated by age of diabetes diagnoses (8.6% of variance explained, p = 0.012). Obesity as measured by several measures, including body mass index (BMI) and measures of central obesity had a negative association with DR; increased measures of obesity consistently reduced odds of DR. As in most earlier studies, DR was associated with the duration of diabetes and glycaemic control. However, other factors, especially obesity related measures were associated with DR, in ways that contrast with most prior studies. These results indicated that the overall pattern of association in the Mumbai slums was novel. Thus, in previously uncharacterized populations, such as the slums that we examined, it is important to evaluate all risk factors de novo to appropriately assess patterns of association as the patterns of association with DR can be complex and population specific.

Hemizygosity can reveal variant pathogenicity on the X-chromosome.

Pathogenic variants on the X-chromosome can have more severe consequences for hemizygous males, while heterozygote females can avoid severe consequences due to diploidy and the capacity for nonrandom expression. Thus, when an allele is more common in females this could indicate that it increases the probability of early death in the male hemizygous state, which can be considered a measure of pathogenicity. Importantly, large-scale genomic data now makes it possible to compare allele proportions between the sexes. To discover pathogenic variants on the X-chromosome, we analyzed exome data from 125,748 ancestrally diverse participants in the Genome Aggregation Database (gnomAD). After filtering out duplicates and extremely rare variants, 44,606 of the original 348,221 remained for analysis. We divided the proportion of variant alleles in females by the proportion in males for all variant sites, and then placed each variant into one of three a priori categories: (1) Reference (Primarily synonymous and intronic), (2) Unlikely-to-be-tolerated (Primarily missense), and (3) Least-likely-to-be-tolerated (Primarily frameshift). To assess the impact of ploidy, we compared the distribution of these ratios between pseudoautosomal and non-pseudoautosomal regions. In the non-pseudoautosomal regions, mean female-to-male ratios were lowest among Reference (2.40), greater for Unlikely-to-be-tolerated (2.77) and highest for Least-likely-to-be-tolerated (3.28) variants. Corresponding ratios were lower in the pseudoautosomal regions (1.52, 1.57, and 1.68, respectively), with the most extreme ratio being just below 11. Because pathogenic effects in the pseudoautosomal regions should not drive ratio increases, this maximum ratio provides an upper bound for baseline noise. In the non-pseudoautosomal regions, 319 variants had a ratio over 11. In sum, we identified a measure with a dataset specific threshold for identifying pathogenicity in non-pseudoautosomal X-chromosome variants: the female-to-male allele proportion ratio.

Methylome-wide Analysis Reveals Epigenetic Marks Associated With Resistance to Tuberculosis in Human Immunodeficiency Virus-Infected Individuals From East Africa.

BACKGROUND: Tuberculosis (TB) is the most deadly infectious disease globally and is highly prevalent in the developing world. For individuals infected with both Mycobacterium tuberculosis (Mtb) and human immunodeficiency virus (HIV), the risk of active TB is 10% or more annually. Previously, we identified in a genome-wide association study (GWAS) a region on chromosome 5 associated with resistance to TB, which included epigenetic marks that could influence gene regulation. We hypothesized that HIV-infected individuals exposed to Mtb who remain disease free carry epigenetic changes that strongly protect them from active TB. METHODS: We conducted a methylome-wide study in HIV-infected, TB-exposed cohorts from Uganda and Tanzania and integrated data from our GWAS. RESULTS: We identified 3 regions of interest that included markers that were differentially methylated between TB cases and controls with latent TB infection: chromosome 1 (RNF220, P = 4 × 10-5), chromosome 2 (between COPS8 and COL6A3, P = 2.7 × 10-5), and chromosome 5 (CEP72, P = 1.3 × 10-5). These methylation results co-localized with associated single-nucleotide polymorphisms (SNPs), methylation QTLs, and methylation × SNP interaction effects. These markers were in regions with regulatory markers for cells involved in TB immunity and/or lung. CONCLUSIONS: Epigenetic regulation is a potential biologic factor underlying resistance to TB in immunocompromised individuals that can act in conjunction with genetic variants.

A Novel Mapping Strategy Utilizing Mouse Chromosome Substitution Strains Identifies Multiple Epistatic Interactions That Regulate Complex Traits.

The genetic contribution of additive vs. non-additive (epistatic) effects in the regulation of complex traits is unclear. While genome-wide association studies typically ignore gene-gene interactions, in part because of the lack of statistical power for detecting them, mouse chromosome substitution strains (CSSs) represent an alternate approach for detecting epistasis given their limited allelic variation. Therefore, we utilized CSSs to identify and map both additive and epistatic loci that regulate a range of hematologic- and metabolism-related traits, as well as hepatic gene expression. Quantitative trait loci (QTL) were identified using a CSS-based backcross strategy involving the segregation of variants on the A/J-derived substituted chromosomes 4 and 6 on an otherwise C57BL/6J genetic background. In the liver transcriptomes of offspring from this cross, we identified and mapped additive QTL regulating the hepatic expression of 768 genes, and epistatic QTL pairs for 519 genes. Similarly, we identified additive QTL for fat pad weight, platelets, and the percentage of granulocytes in blood, as well as epistatic QTL pairs controlling the percentage of lymphocytes in blood and red cell distribution width. The variance attributed to the epistatic QTL pairs was approximately equal to that of the additive QTL; however, the SNPs in the epistatic QTL pairs that accounted for the largest variances were undetected in our single locus association analyses. These findings highlight the need to account for epistasis in association studies, and more broadly demonstrate the importance of identifying genetic interactions to understand the complete genetic architecture of complex traits.

Interaction between host genes and Mycobacterium tuberculosis lineage can affect tuberculosis severity: Evidence for coevolution?

Genetic studies of both the human host and Mycobacterium tuberculosis (MTB) demonstrate independent association with tuberculosis (TB) risk. However, neither explains a large portion of disease risk or severity. Based on studies in other infectious diseases and animal models of TB, we hypothesized that the genomes of the two interact to modulate risk of developing active TB or increasing the severity of disease, when present. We examined this hypothesis in our TB household contact study in Kampala, Uganda, in which there were 3 MTB lineages of which L4-Ugandan (L4.6) is the most recent. TB severity, measured using the Bandim TBscore, was modeled as a function of host SNP genotype, MTB lineage, and their interaction, within two independent cohorts of TB cases, N = 113 and 121. No association was found between lineage and severity, but association between multiple polymorphisms in IL12B and TBscore was replicated in two independent cohorts (most significant rs3212227, combined p = 0.0006), supporting previous associations of IL12B with TB susceptibility. We also observed significant interaction between a single nucleotide polymorphism (SNP) in SLC11A1 and the L4-Ugandan lineage in both cohorts (rs17235409, meta p = 0.0002). Interestingly, the presence of the L4-Uganda lineage in the presence of the ancestral human allele associated with more severe disease. These findings demonstrate that IL12B is associated with severity of TB in addition to susceptibility, and that the association between TB severity and human genetics can be due to an interaction between genes in the two species, consistent with host-pathogen coevolution in TB.

Omega-3 polyunsaturated fatty acid intake norms and preterm birth rate: a cross-sectional analysis of 184 countries.

BACKGROUND: The preponderance of evidence now indicates that elevated long-chain omega-3 polyunsaturated fatty acid (LC omega-3 PUFA) intake is often associated with reduced risk of preterm birth (PTB). This conclusion is based on recent meta-analyses that include several studies that reported null findings. We probed the reasons for this heterogeneity across studies and its implications for PTB prevention using country-level data. METHODS: We analysed the relationship between national PTB rates (<37 weeks of gestation) and omega-3 PUFA intake norms from 184 countries for the year 2010. To estimate the total LC omega-3 PUFA levels (eicosapentaenoic acid [EPA]/docosahexaenoic acid [DHA]) that these norms produce we utilised a metric that accounts for (1) seafood-based omega-3 intake (EPA/DHA) and (2) plant-based omega-3 intake (alpha-linolenic acid [ALA]), ~20% of which is converted to EPA/DHA in vivo. We then assessed the shape of the omega-3-PTB relationship with a penalised spline and conducted linear regression analyses within the linear sections of the relationship. RESULTS: Penalised spline analyses indicated that PTB rates decrease linearly with increasing omega-3 levels up to ~600 mg/day. Income-adjusted linear regression analysis among the countries in this exposure range indicated that the number of PTBs per 100 live births decreases by 1.5 (95% CI 2.8 to 0.3) for each 1 SD increase in omega-3 intake norms (383 mg/day). CONCLUSIONS: Taken with prior evidence for a causal association on the individual level, our findings indicate that omega-3 PUFA deficiency may be a widespread contributing factor in PTB risk. Consideration of baseline omega-3 PUFA levels is critical in the design of future interventions.

The ubiquity of pleiotropy in human disease.

Pleiotropy has long been thought to be a common phenomenon in the human genome; however, until recently appropriate data was unavailable to test this hypothesis. Prior studies focused on assessing the prevalence of pleiotropy in only small subsets of phenotypes (≤ 53 phenotypes), without a truly comprehensive assessment of pleiotropy in the human genome. In this study, we determined the prevalence of pleiotropy, using the entire GWAS catalog (1094 disease phenotypes, 14,459 genes), as well as investigate the relationship between the degree of pleiotropy and the average effect size for each associating gene. The number of associating phenotypes per gene ranged from 1 to 53, with 44% of genes reported in the GWAS catalog associating with more than one phenotype. The proportion of genes shown to be pleiotropic has continued to increase as more studies are added to the catalog. We also found the degree of pleiotropy scales positively with a gene's average effect size (r = 0.04, p value = 0.0003) and negatively with the variance of effect sizes in genes with a given number of associating phenotypes (r = - 0.590, p value = 0.0019). Based on this and prior work, it is becoming evident that pleiotropy is a common, if not ubiquitous, phenomenon. These results have implications in understanding disease etiologies, potentially common biology underlying even disparate diseases, and in elucidating the genotype-phenotype map.

Complex Patterns of Association between Pleiotropy and Transcription Factor Evolution.

Pleiotropy has been claimed to constrain gene evolution but specific mechanisms and extent of these constraints have been difficult to demonstrate. The expansion of molecular data makes it possible to investigate these pleiotropic effects. Few classes of genes have been characterized as intensely as human transcription factors (TFs). We therefore analyzed the evolutionary rates of full TF proteins, along with their DNA binding domains and protein-protein interacting domains (PID) in light of the degree of pleiotropy, measured by the number of TF-TF interactions, or the number of DNA-binding targets. Data were extracted from the ENCODE Chip-Seq dataset, the String v 9.2 database, and the NHGRI GWAS catalog. Evolutionary rates of proteins and domains were calculated using the PAML CodeML package. Our analysis shows that the numbers of TF-TF interactions and DNA binding targets associated with constrained gene evolution; however, the constraint caused by the number of DNA binding targets was restricted to the DNA binding domains, whereas the number of TF-TF interactions constrained the full protein and did so more strongly. Additionally, we found a positive correlation between the number of protein-PIDs and the evolutionary rates of the protein-PIDs. These findings show that not only does pleiotropy associate with constrained protein evolution but the constraint differs by domain function. Finally, we show that GWAS associated TF genes are more highly pleiotropic : The GWAS data illustrates that mutations in highly pleiotropic genes are more likely to be associated with disease phenotypes.

Is Isolated Low High-Density Lipoprotein Cholesterol a Cardiovascular Disease Risk Factor? New Insights From the Framingham Offspring Study.

BACKGROUND: Although the inverse association between high-density lipoprotein cholesterol (HDL-C) and risk of cardiovascular disease (CVD) has been long established, it remains unclear whether low HDL-C remains a CVD risk factor when levels of low-density lipoprotein cholesterol (LDL-C) and triglycerides (TG) are not elevated. This is a timely issue because recent studies have questioned whether HDL-C is truly an independent predictor of CVD. METHODS AND RESULTS: 3590 men and women from the Framingham Heart Study offspring cohort without known CVD were followed between 1987 and 2011. Low HDL-C (<40 mg/dL in men and <50 mg/dL in women) was defined as isolated if TG and LDL-C were both low (<100 mg/dL). We also examined higher thresholds for TG (150 mg/dL) and LDL-C (130 mg/dL) and compared low versus high HDL-C phenotypes using logistic regression analysis to assess association with CVD. Compared with isolated low HDL-C, CVD risks were higher when low HDL-C was accompanied by LDL-C ≥100 mg/dL and TG <100 mg/dL (odds ratio 1.3 [1.0, 1.6]), TG ≥100 mg/dL and LDL-C <100 mg/dL (odds ratio 1.3 [1.1, 1.5]), or TG and LDL-C ≥100 mg/dL (odds ratio 1.6, [1.2, 2.2]), after adjustment for covariates. When low HDL-C was analyzed with higher thresholds for TG (≥150 mg/dL) and LDL-C (≥130 mg/dL), results were essentially the same. In contrast, compared with isolated low HDL-C, high HDL-C was associated with 20% to 40% lower CVD risk except when TG and LDL-C were elevated. CONCLUSIONS: CVD risk as a function of HDL-C phenotypes is modulated by other components of the lipid panel.

Sex-Specific Parental Effects on Offspring Lipid Levels.

BACKGROUND: Plasma lipid levels are highly heritable traits, but known genetic loci can only explain a small portion of their heritability. METHODS AND RESULTS: In this study, we analyzed the role of parental levels of total cholesterol (TC), low-density lipoprotein cholesterol (LDL-C), high-density lipoprotein cholesterol (HDL-C), and triglycerides (TGs) in explaining the values of the corresponding traits in adult offspring. We also evaluated the contribution of nongenetic factors that influence lipid traits (age, body mass index, smoking, medications, and menopause) alone and in combination with variability at the genetic loci known to associate with TC, LDL-C, HDL-C, and TG levels. We performed comparisons among different sex-specific regression models in 416 families from the Framingham Heart Study and 304 from the SardiNIA cohort. Models including parental lipid levels explain significantly more of the trait variation than models without these measures, explaining up to ≈39% of the total trait variation. Of this variation, the parent-of-origin effect explains as much as ≈15% and it does so in a sex-specific way. This observation is not owing to shared environment, given that spouse-pair correlations were negligible (<1.5% explained variation in all cases) and is distinct from previous genetic and acquired factors that are known to influence serum lipid levels. CONCLUSIONS: These findings support the concept that unknown genetic and epigenetic contributors are responsible for most of the heritable component of the plasma lipid phenotype, and that, at present, the clinical utility of knowing age-matched parental lipid levels in assessing risk of dyslipidemia supersedes individual locus effects. Our results support the clinical utility of knowing parental lipid levels in assessing future risk of dyslipidemia.

Diverse convergent evidence in the genetic analysis of complex disease: coordinating omic, informatic, and experimental evidence to better identify and validate risk factors.

In omic research, such as genome wide association studies, researchers seek to repeat their results in other datasets to reduce false positive findings and thus provide evidence for the existence of true associations. Unfortunately this standard validation approach cannot completely eliminate false positive conclusions, and it can also mask many true associations that might otherwise advance our understanding of pathology. These issues beg the question: How can we increase the amount of knowledge gained from high throughput genetic data? To address this challenge, we present an approach that complements standard statistical validation methods by drawing attention to both potential false negative and false positive conclusions, as well as providing broad information for directing future research. The Diverse Convergent Evidence approach (DiCE) we propose integrates information from multiple sources (omics, informatics, and laboratory experiments) to estimate the strength of the available corroborating evidence supporting a given association. This process is designed to yield an evidence metric that has utility when etiologic heterogeneity, variable risk factor frequencies, and a variety of observational data imperfections might lead to false conclusions. We provide proof of principle examples in which DiCE identified strong evidence for associations that have established biological importance, when standard validation methods alone did not provide support. If used as an adjunct to standard validation methods this approach can leverage multiple distinct data types to improve genetic risk factor discovery/validation, promote effective science communication, and guide future research directions.

G6PD A- deficiency and severe malaria in The Gambia: heterozygote advantage and possible homozygote disadvantage.

Glucose-6-phosphate dehydrogenase (G6PD) deficiency is frequent in Africa, because it confers resistance to Plasmodium falciparum malaria; however, the nature of the protection and the genotypes associated with it have been controversial. In 1972, Bienzle and others described protection from malaria in West African females heterozygous for G6PD A-. They determined that G6PD A- heterozygotes had lower parasite counts than A- homozygotes, hemizygous males, and normal individuals. However, other studies have reached different conclusions about the protective genotypes. DNA samples from 135 children with severe malaria and 146 children with mild malaria from The Gambia were genotyped for the G6PD A- mutation that is most frequent among Gambians (G6PD 968 T->C); there was a marked deficiency of heterozygotes and an excess of homozygotes with severe malaria, producing a strong deviation from Hardy-Weinberg equilibrium. Our results support the protective effect in G6PD A- heterozygous females and suggest that homozygotes might be more susceptible to severe malaria attacks.

Increased variance in germline allele-specific expression of APC associates with colorectal cancer.

BACKGROUND & AIMS: Germline variations in allele-specific expression (ASE) are associated with highly penetrant familial cancers, but their role in common sporadic cancers is unclear. ASE of adenomatous polyposis coli (APC) is associated with pathogenesis of familial adenomatous polyposis. We investigated whether moderate variations in ASE of APC contribute to common forms of colorectal cancer (CRC). METHODS: Denaturing high-performance liquid chromatography was used to analyze germline ASE of APC in blood samples from patients with CRC (cases, n = 53) and controls (n = 68). Means, medians, and variances of ASE were compared. Variants in the APC gene region also were analyzed. RESULTS: The distribution of ASE differed significantly between groups; cases had significantly larger amounts of variance than controls (P = .0004). Risk for CRC increased proportionally with the degree of deviation from the mean. The odds ratio for individuals with levels of ASE that deviated more than 1 standard deviation from the mean was 3.97 (95% confidence interval, 1.71-9.24; P = .001); for those with levels greater than 1.645 standard deviations, the odds ratio was 13.46 (95% confidence interval, 1.76-609.40; P = .005). Sequence analysis revealed that a patient with a high level of ASE who did not have a family history of CRC carried a nonsense mutation in APC (p.Arg216X). Genotype analysis of APC associated multiple single-nucleotide polymorphisms with ASE values and/or variance among cases, but not controls. Cis variants, therefore, might account for some of the variance in ASE of APC. CONCLUSIONS: Patients with CRC have a larger variance in germline levels of ASE in APC than controls; large distances from the mean ASE were associated with risk for common forms of CRC.

Polymorphisms in maternal and fetal genes encoding for proteins involved in extracellular matrix metabolism alter the risk for small-for-gestational-age.

OBJECTIVE: To examine the association between maternal and fetal genetic variants and small-for-gestational-age (SGA). METHODS: A case-control study was conducted in patients with SGA neonates (530 maternal and 436 fetal) and controls (599 maternal and 628 fetal); 190 candidate genes and 775 SNPs were studied. Single-locus, multi-locus and haplotype association analyses were performed on maternal and fetal data with logistic regression, multifactor dimensionality reduction (MDR) analysis, and haplotype-based association with 2 and 3 marker sliding windows, respectively. Ingenuity pathway analysis (IPA) software was used to assess pathways that associate with SGA. RESULTS: The most significant single-locus association in maternal data was with a SNP in tissue inhibitor of metalloproteinase 2 (TIMP2) (rs2277698 OR = 1.71, 95% CI [1.26-2.32], p = 0.0006) while in the fetus it was with a SNP in fibronectin 1 isoform 3 preproprotein (FN1) (rs3796123, OR = 1.46, 95% CI [1.20-1.78], p = 0.0001). Both SNPs were adjusted for potential confounders (maternal body mass index and fetal sex). Haplotype analyses resulted in associations in α 1 type I collagen preproprotein (COL1A1, rs1007086-rs2141279-rs17639446, global p = 0.006) in mothers and FN1 (rs2304573-rs1250204-rs1250215, global p = 0.045) in fetuses. Multi-locus analyses with MDR identified a two SNP model with maternal variants collagen type V α 2 (COL5A2) and plasminogen activator urokinase (PLAU) predicting SGA outcome correctly 59% of the time (p = 0.035). CONCLUSIONS: Genetic variants in extracellular matrix-related genes showed significant single-locus association with SGA. These data are consistent with other studies that have observed elevated circulating fibronectin concentrations in association with increased risk of SGA. The present study supports the hypothesis that DNA variants can partially explain the risk of SGA in a cohort of Hispanic women.

A genetic association study of maternal and fetal candidate genes that predispose to preterm prelabor rupture of membranes (PROM).

OBJECTIVE: We sought to determine whether maternal/fetal single-nucleotide polymorphisms (SNPs) in candidate genes are associated with preterm prelabor rupture of membranes (pPROM). STUDY DESIGN: A case-control study was conducted in patients with pPROM (225 mothers and 155 fetuses) and 599 mothers and 628 fetuses with a normal pregnancy; 190 candidate genes and 775 SNPs were studied. Single locus/haplotype association analyses were performed; false discovery rate was used to correct for multiple testing (q* = 0.15). RESULTS: First, a SNP in tissue inhibitor of metalloproteinase 2 in mothers was significantly associated with pPROM (odds ratio, 2.12; 95% confidence interval, 1.47-3.07; P = .000068), and this association remained significant after correction for multiple comparisons. Second, haplotypes for Alpha 3 type IV collagen isoform precursor in the mother were associated with pPROM (global P = .003). Third, multilocus analysis identified a 3-locus model, which included maternal SNPs in collagen type I alpha 2, defensin alpha 5 gene, and endothelin 1. CONCLUSION: DNA variants in a maternal gene involved in extracellular matrix metabolism doubled the risk of pPROM.

Identification of fetal and maternal single nucleotide polymorphisms in candidate genes that predispose to spontaneous preterm labor with intact membranes.

OBJECTIVE: The purpose of this study was to determine whether maternal/fetal single nucleotide polymorphisms (SNPs) in candidate genes are associated with spontaneous preterm labor/delivery. STUDY DESIGN: A genetic association study was conducted in 223 mothers and 179 fetuses (preterm labor with intact membranes who delivered <37 weeks of gestation [preterm birth (PTB)]), and 599 mothers and 628 fetuses (normal pregnancy); 190 candidate genes and 775 SNPs were studied. Single locus/haplotype association analyses were performed; the false discovery rate was used to correct for multiple testing. RESULTS: The strongest single locus associations with PTB were interleukin-6 receptor 1 (fetus; P=.000148) and tissue inhibitor of metalloproteinase 2 (mother; P=.000197), which remained significant after correction for multiple comparisons. Global haplotype analysis indicated an association between a fetal DNA variant in insulin-like growth factor F2 and maternal alpha 3 type IV collagen isoform 1 (global, P=.004 and .007, respectively). CONCLUSION: An SNP involved in controlling fetal inflammation (interleukin-6 receptor 1) and DNA variants in maternal genes encoding for proteins involved in extracellular matrix metabolism approximately doubled the risk of PTB.

A prevalence-based association test for case-control studies.

Genetic association is often determined in case-control studies by the differential distribution of alleles or genotypes. Recent work has demonstrated that association can also be assessed by deviations from the expected distributions of alleles or genotypes. Specifically, multiple methods motivated by the principles of Hardy-Weinberg equilibrium (HWE) have been developed. However, these methods do not take into account many of the assumptions of HWE. Therefore, we have developed a prevalence-based association test (PRAT) as an alternative method for detecting association in case-control studies. This method, also motivated by the principles of HWE, uses an estimated population allele frequency to generate expected genotype frequencies instead of using the case and control frequencies separately. Our method often has greater power, under a wide variety of genetic models, to detect association than genotypic, allelic or Cochran-Armitage trend association tests. Therefore, we propose PRAT as a powerful alternative method of testing for association.

Discussing gene-gene interaction: warning--translating equations to English may result in jabberwocky.

Interest in mapping susceptibility alleles for complex diseases, which do not follow a classic single-gene segregation pattern, has driven interest in methods that account for, or use information from one locus when mapping another. Our discussion group examined methods related to epistasis or gene x gene interaction. The goal of modeling gene x gene interaction varied across groups; some papers tried to detect gene x gene interaction while others tried to exploit it to map genes. Most of the 10 papers summarized here applied newly created or newly modified statistical methods related to gene x gene interaction, while two groups primarily examined computational issues. As is often the case, comparisons are complicated by little overlap in the data used across the papers, and further complicated by the fact that the available data may not have been ideal for some gene x gene interaction methods. However, the main difficulty in comparing and contrasting methods across the papers is the lack of a consistent statistical definition of gene x gene interaction. But despite these issues, two clear trends emerged across the analyses: First, the methods for quantitative trait gene x gene interaction appeared to perform very well, even in families initially ascertained as affected sib pairs; and second, dichotomous trait gene x gene interaction methods failed to produce consistent results. The difficulty of using (primarily) affected sib pair data in a gene x gene interaction analysis is explored.

Exploring epistasis in candidate genes for rheumatoid arthritis.

The identification of susceptibility genes for common, chronic disease presents great challenges. The development of novel statistical and computational methodologies to help identify these genes is an area of great necessity. Much research is ongoing and the Genetic Analysis Workshop (GAW) is a venue for the dissemination and comparison of many of these methods. GAW15 included real data sets to look for disease susceptibility genes for rheumatoid arthritis (RA). RA is a complex, chronic inflammatory disease with several replicated disease genes, but much of the genetic variation in the phenotype remains unexplained. We applied two computational methods, namely multifactor dimensionality reduction (MDR) and grammatical evolution neural networks (GENN), to three data sets from GAW15. While these analytic methods were applied with the intention of detecting of multilocus models of association, both methods identified a strong single locus effect of a single-nucleotide polymorphism (SNP) in PTPN22 that is significantly associated with RA. This SNP has previously been associated with RA in several other published studies. These results demonstrate that both MDR and GENN are capable of identifying a single-locus main effect, in addition to multilocus models of association. This is the first published comparison of the two methods. Because GENN employs an evolutionary computation search strategy in comparison to the exhaustive search strategy of MDR, it is encouraging that the two methods produced similar results. This comparison should be extended in future studies with both simulated and real data.