Asymptotics for Marginal Generalized Linear Models With Sparse Correlations

Marginal generalized linear models can be used for clustered and longitudinal data by fitting a model as if the data were independent and using an empirical estimator of parameter standard errors. We extend this approach to data where the number of observations correlated with a given one grows with sample size and show that parameter estimates are consistent and asymptotically Normal with a slower convergence rate than for independent data, and that an information sandwich variance estimator is consistent. We present two problems that motivated this work, the modelling of patterns of HIV genetic variation and the behavior of clustered data estimators when clusters are large.

Ecotypes and genetic structure of Rhinanthus alectorolophus (Orobanchaceae) in southwestern Germany

European annual species of the genus Rhinanthus often exhibit seasonal ecotypic variation, a phenomenon also known from related genera of hemiparasitic Orobanchaceae. Populations with different flowering times exist, correlated with differences in a number of morphological characters. The present study evaluates the correlation of morphological characters and genetic differentiation of populations of Rhinanthus alectorolophus. Thirty-nine populations of three different subspecies from southwestern Germany were sampled. A total of 798 individuals were used for morphological analyses and 187 of these for AFLP analyses. Principal component analysis showed that morphological variation is mostly continuous. In a discriminant analysis based on morphological characters, only 89.7 % of all individuals were correctly assigned to their previously determined subspecies, indicating that subspecies identification is ambiguous for some populations. Using AFLP data and Bayesian assignment analysis, the sampled individuals could be grouped in three genetic clusters which do not correspond to the three subspecies. Instead, the clustering shows a clear geographic pattern and a Mantel test likewise revealed a significant correlation between genetic and geographic distances. Correlations of genetic distances with differences in morphological characters were weak and mostly insignificant. The results indicate that the subspecies of R. alectorolophus do not form discrete entities and that the character combinations distinguishing them are homoplastic.

Temporal genetic structure and relatedness in the Tufted Duck Aythya fuligula suggests limited kin association in winter

Conspecific aggregation of waterfowl in winter is a common example of animal flocking behaviour, yet patterns of relatedness and temporal substructure in such social groups remain poorly understood even in common species. A previous study based on mark-recapture data showed that Tufted Ducks Aythya fuligula caught on the same day were re-caught together in subsequent winters more often than expected by chance, suggesting stable assortments of ‘socially familiar’ individuals between wintering periods. The genetic relationships within these social groups were not clear. Based on 191 individuals genotyped at 10 microsatellite markers, we investigated the temporal genetic structure and patterns of relatedness among wintering Tufted Ducks at Lake Sempach, Switzerland, in two consecutive winters. We found no evidence of genetic differentiation between temporal groups within or between winters. The average levels of relatedness in temporal groups were low and not higher than expected in random assortments of individuals. However, Mantel tests performed for each sex separately revealed significant negative correlations between the pairwise relatedness coefficients and the number of days between the capture dates of pairs of wintering Tufted Duck in males and females. This pattern suggests the presence of a small number of co-migrating same-sex sibling pairs in wintering flocks of Tufted Ducks. Our findings provide one of the first genetic analyses of a common duck species outside the breeding season and contribute to the understanding of social interactions in long-distance migratory birds.

The interplay between host genetic variation, viral replication and microbial translocation in untreated HIV-infected individuals.

Systemic immune activation, a major determinant of HIV disease progression, is the result of a complex interplay between viral replication, dysregulation of the immune system, and microbial translocation due to gut mucosal damage. While human genetic variants influencing HIV viral load have been identified, it is unknown to what extent the host genetic background contributes to inter-individual differences in other determinants of HIV pathogenesis like gut damage and microbial translocation. Using samples and data from 717 untreated participants in the Swiss HIV Cohort Study and a genome-wide association study design, we searched for human genetic determinants of plasma levels of intestinal fatty-acid binding protein (I-FABP/FABP2), a marker of gut damage, and of soluble sCD14 (sCD14), a marker of LPS bioactivity and microbial translocation. We also assessed the correlations between HIV viral load, sCD14 and I-FABP. While we found no genome-wide significant determinant of the tested plasma markers, we observed strong associations between sCD14 and both HIV viral load and I-FABP, shedding new light on the relationships between processes that drive progression of untreated HIV infection.

Expansion load: recessive mutations and the role of standing genetic variation

Expanding populations incur a mutation burden – the so-called expansion load. Previous studies of expansion load have focused on codominant mutations. An important consequence of this assumption is that expansion load stems exclusively from the accumulation of new mutations occurring in individuals living at the wave front. Using individual-based simulations, we study here the dynamics of standing genetic variation at the front of expansions, and its consequences on mean fitness if mutations are recessive. We find that deleterious genetic diversity is quickly lost at the front of the expansion, but the loss of deleterious mutations at some loci is compensated by an increase of their frequencies at other loci. The frequency of deleterious homozygotes therefore increases along the expansion axis, whereas the average number of deleterious mutations per individual remains nearly constant across the species range. This reveals two important differences to codominant models: (i) mean fitness at the front of the expansion drops much faster if mutations are recessive, and (ii) mutation load can increase during the expansion even if the total number of deleterious mutations per individual remains constant. We use our model to make predictions about the shape of the site frequency spectrum at the front of range expansion, and about correlations between heterozygosity and fitness in different parts of the species range. Importantly, these predictions provide opportunities to empirically validate our theoretical results. We discuss our findings in the light of recent results on the distribution of deleterious genetic variation across human populations and link them to empirical results on the correlation of heterozygosity and fitness found in many natural range expansions.

Bayesian models for the analysis of genetic structure when populations are correlated

Motivation: Population allele frequencies are correlated when populations have a shared history or when they exchange genes. Unfortunately, most models for allele frequency and inference about population structure ignore this correlation. Recent analytical results show that among populations, correlations can be very high, which could affect estimates of population genetic structure. In this study, we propose a mixture beta model to characterize the allele frequency distribution among populations. This formulation incorporates the correlation among populations as well as extending the model to data with different clusters of populations. Results: Using simulated data, we show that in general, the mixture model provides a good approximation of the among-population allele frequency distribution and a good estimate of correlation among populations. Results from fitting the mixture model to a dataset of genotypes at 377 autosomal microsatellite loci from human populations indicate high correlation among populations, which may not be appropriate to neglect. Traditional measures of population structure tend to over-estimate the amount of genetic differentiation when correlation is neglected. Inference is performed in a Bayesian framework.

Importance of epistasis as the genetic basis of heterosis in an elite rice hybrid

The genetic basis of heterosis was investigated in an elite rice hybrid by using a molecular linkage map with 150 segregating loci covering the entire rice genome. Data for yield and three traits that were components of yield were collected over 2 years from replicated field trials of 250 F2:3 families. Genotypic variations explained from about 50% to more than 80% of the total variation. Interactions between genotypes and years were small compared with the main effects. A total of 32 quantitative trait loci (QTLs) were detected for the four traits; 12 were observed in both years and the remaining 20 were detected in only one year. Overdominance was observed for most of the QTLs for yield and also for a few QTLs for the component traits. Correlations between marker heterozygosity and trait expression were low, indicating that the overall heterozygosity made little contribution to heterosis. Digenic interactions, including additive by additive, additive by dominance, and dominance by dominance, were frequent and widespread in this population. The interactions involved large numbers of marker loci, most of which individually were not detectable on single-locus basis; many interactions among loci were detected in both years. The results provide strong evidence that epistasis plays a major role as the genetic basis of heterosis.

Gene-based approach to human gene-phenotype correlations

Elucidating the genetic basis of human phenotypes is a major goal of contemporary geneticists. Logically, two fundamental and contrasting approaches are available, one that begins with a phenotype and concludes with the identification of a responsible gene or genes; the other that begins with a gene and works toward identifying one or more phenotypes resulting from allelic variation of it. This paper provides a conceptual overview of phenotype-based vs. gene-based procedures with emphasis on gene-based methods. A key feature of a gene-based approach is that laboratory effort first is devoted to developing an assay for mutations in the gene under regard; the assay then is applied to the evaluation of large numbers of unrelated individuals with a variety of phenotypes that are deemed potentially resulting from alleles at the gene. No effort is directed toward chromosomally mapping the loci responsible for the phenotypes scanned. Example is made of my laboratory’s successful use of a gene-based approach to identify genes causing hereditary diseases of the retina such as retinitis pigmentosa. Reductions in the cost and improvements in the speed of scanning individuals for DNA sequence anomalies may make a gene-based approach an efficient alternative to phenotype-based approaches to correlating genes with phenotypes.

Molecular Genetic Basis of Opposite Gene by Environment Interactions in Reading Disability and Attention Deficit/Hyperactivity Disorder

The goal of this study is to better understand the genetic basis of Reading Disability (RD) and Attention Deficit Hyperactivity Disorder (ADHD) by examining molecular G x E interactions with parental education for each disorder. Research indicates that despite sharing genetic risk factors, RD and ADHD are influenced by different types of G x E interactions with parental education - a diathesis stress interaction in the case of ADHD and a bioecological interaction in RD. In order to resolve this apparent paradox, we conducted a preliminary study using behavioral genetic methods to test for G x E interactions in RD and the inattentive subtype of ADHD (ADHD-I) in the same sample of monozygotic and dizygotic Colorado Learning Disabilities Research Center same-sex twin pairs (DeFries et al., 1997), and our findings were consistent with the literature. We posited a genetic hypothesis for this opposite pattern of interactions, which suggests that only genes specific to each disorder enter into these opposite interactions, not the shared genes underlying their comorbidity. This study sought to further investigate this paradox using molecular genetics methods. We examined multiple candidate genes identified for RD or related language phenotypes and those identified for ADHD for G x E interactions with parental education. The specific aims of this study were as follows: 1) partition known risk alleles for RD and/or related language phenotypes and ADHD-I into those which are pleiotropic and non-pleiotropic by testing each risk allele for association with both RD and ADHD-I, 2) explore the main effects of parental education on both RD and ADHD-I, 3) address G-E correlations, and 4) conduct exploratory G x E interaction analyses in order to test the genetic hypothesis. Analyses suggested a number of pleiotropic genes that influence both RD and ADHD; however, results did not remain after correcting for multiple comparisons. Although exploratory G x E interaction findings were not significant after multiple comparison correction, results suggested a G x E interaction in the bioecological direction with KIAA0319, parental education, and ADHD-I. Given the limited power in the current study, replication of these findings with larger samples is necessary.

A genetic two-factor model of the covariation among a subset of Multidimensional Aptitude Battery and Wechsler Adult Intelligence Scale - Revised subtests

The phenotypic and genetic factor structure of performance on five Multidimensional Aptitude Battery (MAB) subtests and one Wechsler Adult Intelligence Scale-Revised (WAIS-R) subtest was explored in 390 adolescent twin pairs (184 monozygotic [MZ]; 206 dizygotic (DZ)). The temporal stability of these measures was derived from a subsample of 49 twin pairs, with test-retest correlations ranging from .67 to .85. A phenotypic factor model, in which performance and verbal factors were correlated, provided a good fit to the data. Genetic modeling was based on the phenotypic factor structure, but also took into account the additive genetic (A), common environmental (C), and unique environmental (E) parameters derived from a fully saturated ACE model. The best fitting model was characterized by a genetic correlated two-factor structure with specific effects, a general common environmental factor, and overlapping unique environmental effects. Results are compared to multivariate genetic models reported in children and adults, with the most notable difference being the growing importance of common genes influencing diverse abilities in adolescence. (C) 2003 Elsevier Inc. All rights reserved.

Genetic and environmental sources of covariance between reading tests used in neuropsychological assessment and IQ subtests

In this study, we examined genetic and environmental influences on covariation among two reading tests used in neuropsychological assessment (Cambridge Contextual Reading Test [CCRT], [Beardsall, L., and Huppert, F. A. ( 1994). J. Clin. Exp. Neuropsychol. 16: 232 - 242], Schonell Graded Word Reading Test [SGWRT], [ Schonell, F. J., and Schonell, P. E. ( 1960). Diagnostic and attainment testing. Edinburgh: Oliver and Boyd.]) and among a selection of IQ subtests from the Multidimensional Aptitude Battery (MAB), [Jackson, D. N. (1984). Multidimensional aptitude battery, Ontario: Research Psychologists Press.] and the Wechsler Adult Intelligence Scale-Revised (WAIS-R) [Wechsler, D. (1981). Manual for the Wechsler Adult Intelligence Scale-Revised (WAIS-R). San Antonio: The Psychological Corporation]. Participants were 225 monozygotic and 275 dizygotic twin pairs aged from 15 years to 18 years ( mean, 16 years). For Verbal IQ subtests, phenotypic correlations with the reading tests ranged from 0.44 to 0.65. For Performance IQ subtests, phenotypic correlations with the reading tests ranged from 0.23 to 0.34. Results of Structural Equation Modeling (SEM) supported a model with one genetic General factor and three genetic group factors ( Verbal, Performance, Reading). Reading performance was influenced by the genetic General factor ( accounting for 13% and 20% of the variance for the CCRT and SGWRT, respectively), the genetic Verbal factor ( explaining 17% and 19% of variance for the CCRT and SGWRT), and the genetic Reading factor ( explaining 21% of the variance for both the CCRT and SGWRT). A common environment factor accounted for 25% and 14% of the CCRT and SGWRT variance, respectively. Genetic influences accounted for more than half of the phenotypic covariance between the reading tests and each of the IQ subtests. The heritabilities of the CCRT and SGWRT were 0.54 and 0.65, respectively. Observable covariance between reading assessments used by neuropsychologists to estimate IQ and IQ subtests appears to be largely due to genetic effects.

The genetic basis of academic achievement on the Queensland Core Skills Test and its shared genetic variance with IQ

First, this study examined genetic and environmental sources of variation in performance on a standardised test of academic achievement, the Queensland Core Skills Test (QCST) (Queensland Studies Authority, 2003a). Second, it assessed the genetic correlation among the QCST score and Verbal and Performance IQ measures using the Multidimensional Aptitude Battery (MAB), [Jackson, D. N. (1984) Multidimensional Aptitude Battery manual. Port Huron, MI:Research Psychologist Press, Inc.]. Participants were 256 monozygotic twin pairs and 326 dizygotic twin pairs aged from 15 to 18 years (mean 17 years +/- 0.4 [SD]) when achievement tested, and from 15 to 22 years (mean 16 years +/- 0.4 [SD]) when IQ tested. Univariate analysis indicated a heritability for the QCST of 0.72. Adjustment to this estimate due to truncate selection (downward adjustment) and positive phenotypic assortative mating (upward adjustment) suggested a heritability of 0.76 The phenotypic (0.81) and genetic (0.91) correlations between the QCST and Verbal IQ (VIQ) were significantly stronger than the phenotypic (0.57) and genetic (0.64) correlations between the QCST and Performance IQ (PIQ). The findings suggest that individual variation in QCST performance is largely due to genetic factors and that common environmental effects may be substantially accounted for by phenotypic assortative mating. Covariance between academic achievement on the QCST and psychometric IQ (particularly VIQ) is to a large extent due to common genetic influences.

Genetic and Cultural Transmission of Smoking Initiation: An Extended Twin Kinship Model

Background Considerable evidence from twin and adoption studies indicates that genetic and shared environmental factors play a significant role in the initiation of smoking behavior. Although twin and adoption designs are powerful to detect genetic and environmental influences, they do not provide information on the processes of assortative mating and parent–offspring transmission and their contribution to the variability explained by genetic and/or environmental factors. Methods We examined the role of genetic and environmental factors for smoking initiation using an extended kinship design. This design allows the simultaneous testing of additive and non-additive genetic, shared and individual-specific environmental factors, as well as sex differences in the expression of genes and environment in the presence of assortative mating and combined genetic and cultural transmission. A dichotomous lifetime smoking measure was obtained from twins and relatives in the Virginia 30,000 sample. Results Results demonstrate that both genetic and environmental factors play a significant role in the liability to smoking initiation. Major influences on individual differences appeared to be additive genetic and unique environmental effects, with smaller contributions from assortative mating, shared sibling environment, twin environment, cultural transmission and resulting genotype–environment covariance. The finding of negative cultural transmission without dominance led us to investigate more closely two possible mechanisms for the lower parent–offspring correlations compared to the sibling and DZ twin correlations in subsets of the data: (i) age × gene interaction, and (ii) social homogamy. Neither mechanism provided a significantly better explanation of the data, although age regression was significant. Conclusions This study showed significant heritability, partly due to assortment, and significant effects of primarily non-parental shared environment on smoking initiation.

Genetic markers, birth characteristics, and childhood leukemia risk

The cause for childhood acute lymphoblastic leukemia (ALL) remains unknown, but male gender is a risk factor, and among ethnicities, Hispanics have the highest risk. In this dissertation, we explored correlations among genetic polymorphisms, birth characteristics, and the risk of childhood ALL in a multi-ethnic sample in 161 cases and 231 controls recruited contemporaneously (2007-2012) in Houston, TX. We first examined three lymphoma risk markers, since lymphoma and ALL both stem from lymphoid cells. Of these, rs2395185 showed a risk association in non-Hispanic White males (OR=2.8, P=0.02; P interaction=0.03 for gender), but not in Hispanics. We verified previously known risk associations to validate the case-control sample. Mutations of HFE (C282Y, H63D) were genotyped to test whether iron-regulatory gene (IRG) variants known to elevate iron levels increase childhood ALL risk. Being positive for either polymorphism yielded only a modestly elevated OR in males, which increased to 2.96 (P=0.01) in the presence of a particular transferrin receptor (TFRC) genotype for rs3817672 (Pinteraction=0.04). SNP rs3817672 itself showed an ethnicity-specific association (P interaction=0.02 for ethnicity). We then examined additional IRG SNPs (rs422982, rs855791, rs733655), which showed risk associations in males (ORs=1.52 to 2.60). A polygenic model based on the number of polymorphic alleles in five IRG SNPs revealed a linear increase in risk (OR=2.00 per incremental change; P=0.002). Having three or more alleles compared with none was associated with increased risk in males (OR=4.12; P=0.004). Significant risk associations with childhood ALL was found with birth length (OR=1.18 per inch, P=0.04), high birth weight (>4,000g) (OR=1.93, P=0.01), and with gestational age (OR=1.10 per week, P=0.04). We observed a negative correlation between HFE SNP rs9366637 and gestational age (P=0.005), again, stronger in males ( P=0.001) and interacting with TFRC (P interaction=0.05). Our results showed that (i) ALL risk markers do not show universal associations across ethnicities or between genders, (ii) IRG SNPs modify ALL risk presumably by their effects on iron levels, (iii) a negative correlation between an HFE SNP and gestational age exists, which implicates an iron-related mechanism. The results suggest that currently unregulated supplemental iron intake may have implications on childhood ALL development.

Peroxisome Proliferator-Activated Receptor-γ Coactivator 1-α (PPARGC1A) Genetic Associations with Type 2 Diabetes in Three Ethnicities

Genetic heterogeneity, lifestyle factors, gene-gene or gene-environment interactions are the determinants of T2D which puts Hispanics and populations with African ancestry at higher risk of developing T2D. In this dissertation, the genetic associations of PPARGC1A polymorphisms with T2D and its related phenotypes (metabolic markers) in Haitian Americans (cases=110, controls=116), African Americans (cases=120, controls=124) and Cuban Americans (cases=160, controls=181) of South Florida were explored. Five single nucleotide polymorphisms of gene PPARGC1A were evaluated in each ethnicity for their disease association. In Haitian Americans, rs7656250 (OR= 0.22, pp=0.03) had significant protective association with T2D but had risk association in African Americans for rs7656250 (OR=1.02, p=0.96) and rs4235308 (OR=2.53, p=0.03). We found that in Haitian American females, both rs7656250 (OR=0.23, pp=0.03) had protective association with T2D. In African American females, rs7656250 (OR=1.14, p=0.78) had risk association whereas in males, it had significant protective effect (OR=0.37, p=0.04). However, the risk association exhibited by rs4235308 was stronger in African American females (OR=2.69, p=0.03) than males (OR=1.16, p=0.72). In Cuban Americans, only rs7656250 showed significant risk association with T2D (OR=6.87, p=0.02) which was stronger in females alone (OR=7.67, p=0.01). We also observed significant differences among correlations of PPARGC1A SNPs and T2D phenotypes. Positive correlation was observed for log Hs-CRP with rs3774907 (pp=0.03) in Cuban Americans respectively. Correlation of log A1C with rs7656250 (p=0.02) was positive in Cuban Americans while it was negative for rs3774907 in Haitian Americans (ppPPARGC1A correlations with T2D and its phenotypes among the three ethnicities studied (ii) the associations of PPARGC1A SNPs showed significant effect modification by sex. The findings suggest that variations in effects of PPARGC1A gene polymorphisms among three ethnicities and between sexes may have biomedical implications for the development of T2D as well as the phenotypes related to T2D.