Locating the genes underlying a simulated complex disease by discriminant analysis

Analysis of heterogeneity and epistasis in physiological mixed populations by combined structural equation modelling and latent class analysis

BACKGROUND

Biological systems are interacting, molecular networks in which genetic variation contributes to phenotypic heterogeneity. This heterogeneity is traditionally modelled as a dichotomous trait (e.g. affected vs. non-affected). This is far too simplistic considering the complexity and genetic variations of such networks.

METHODS

In this study on type 2 diabetes mellitus, heterogeneity was resolved in a latent class framework combined with structural equation modelling using phenotypic indicators of distinct physiological processes. We modelled the clinical condition "the metabolic syndrome", which is known to be a heterogeneous and polygenic condition with a clinical endpoint (type 2 diabetes mellitus). In the model presented here, genetic factors were not included and no genetic model is assumed except that genes operate in networks. The impact of stratification of the study population on genetic interaction was demonstrated by analysis of several genes previously associated with the metabolic syndrome and type 2 diabetes mellitus.

RESULTS

The analysis revealed the existence of 19 distinct subpopulations with a different propensity to develop diabetes mellitus within a large healthy study population. The allocation of subjects into subpopulations was highly accurate with an entropy measure of nearly 0.9. Although very few gene variants were directly associated with metabolic syndrome in the total study sample, almost one third of all possible epistatic interactions were highly significant. In particular, the number of interactions increased after stratifying the study population, suggesting that interactions are masked in heterogenous populations. In addition, the genetic variance increased by an average of 35-fold when analysed in the subpopulations.

CONCLUSION

The major conclusions from this study are that the likelihood of detecting true association between genetic variants and complex traits increases tremendously when studied in physiological homogenous subpopulations and on inclusion of epistasis in the analysis, whereas epistasis (i.e. genetic networks) is ubiquitous and should be the basis in modelling any biological process.

Multi-Locus Penetrance Variance Analysis Method for Association Study in Complex Diseases

Common heritable diseases often result from the action of several different genes, each of which contributes to the total observed variability in the disease trait. Traditional single-locus association approaches rely heavily on the marginal effects of single-locus and tend to ignore the multigenic nature of complex diseases. The increasing request for localizing genes underlying traits in multi-gene diseases has led to the development of some statistical methods. In this study, we develop a multi-locus analysis method - multi-locus penetrance variance analysis (MPVA), and conduct systematical simulation studies to evaluate its performance. Our results show that compared with other multi-locus methods, MPVA has some advantage in detecting complicated interactions under different epistatic models, and its performance is stable and robust.

Multi-locus association study of schizophrenia susceptibility genes with a posterior probability method

Schizophrenia is a serious neuropsychiatric illness affecting about 1% of the world's population. It is considered a complex inheritance disorder. A number of genes are involved in combination in the etiology of the disorder. Evidence implicates the altered dopaminergic transmission in schizophrenia. In the present study, in order to identify susceptibility genes for schizophrenia in dopaminergic metabolism, we analyzed 59 single nucleotide polymorphisms (SNPs) in 24 genes of the dopaminergic pathway among 82 unrelated patients with schizophrenia and 108 matched normal controls. Considering that traditional single-locus association studies ignore the multigenic nature of complex diseases and do not take into account possible interactions between susceptibility genes, we proposed a multi-locus analysis method, using the posterior probability of morbidity as a measure of absolute disease risk for a multi-locus genotype combination, and developed an algorithm based on perturbation and average to detect the susceptibility multi-locus genotype combinations, as well as to repress noise and avoid false positive results at our best. A three-locus SNP genotype combination involved in the interactions of COMT and ALDH3B1 genes was detected to be significantly susceptible to schizophrenia.

Data mining and computationally intensive methods: summary of Group 7 contributions to Genetic Analysis Workshop 13

The Framingham Heart Study data, as well as a related simulated data set, were generously provided to the participants of the Genetic Analysis Workshop 13 in order that newly developed and emerging statistical methodologies could be tested on that well-characterized data set. The impetus driving the development of novel methods is to elucidate the contributions of genes, environment, and interactions between and among them, as well as to allow comparison between and validation of methods. The seven papers that comprise this group used data-mining methodologies (tree-based methods, neural networks, discriminant analysis, and Bayesian variable selection) in an attempt to identify the underlying genetics of cardiovascular disease and related traits in the presence of environmental and genetic covariates. Data-mining strategies are gaining popularity because they are extremely flexible and may have greater efficiency and potential in identifying the factors involved in complex disorders. While the methods grouped together here constitute a diverse collection, some papers asked similar questions with very different methods, while others used the same underlying methodology to ask very different questions. This paper briefly describes the data-mining methodologies applied to the Genetic Analysis Workshop 13 data sets and the results of those investigations.

Multivariate sib-pair linkage analysis of longitudinal phenotypes by three step-wise analysis approaches

Background

Current statistical methods for sib-pair linkage analysis of complex diseases include linear models, generalized linear models, and novel data mining techniques. The purpose of this study was to further investigate the utility and properties of a novel pattern recognition technique (step-wise discriminant analysis) using the chromosome 10 linkage data from the Framingham Heart Study and by comparing it with step-wise logistic regression and linear regression.

Results

The three step-wise approaches were compared in terms of statistical significance and gene localization. Step-wise discriminant linkage analysis approach performed best; next was step-wise logistic regression; and step-wise linear regression was the least efficient because it ignored the categorical nature of disease phenotypes. Nevertheless, all three methods successfully identified the previously reported chromosomal region linked to human hypertension, marker GATA64A09. We also explored the possibility of using the discriminant analysis to detect gene × gene and gene × environment interactions. There was evidence to suggest the existence of gene × environment interactions between markers GATA64A09 or GATA115E01 and hypertension treatment and gene × gene interactions between markers GATA64A09 and GATA115E01. Finally, we answered the theoretical question "Is a trichotomous phenotype more efficient than a binary?" Unlike logistic regression, discriminant sib-pair linkage analysis might have more power to detect linkage to a binary phenotype than a trichotomous one.

Conclusion

We confirmed our previous speculation that step-wise discriminant analysis is useful for genetic mapping of complex diseases. This analysis also supported the possibility of the pattern recognition technique for investigating gene × gene or gene × environment interactions.

Genes and gene polymorphisms associated with periodontal disease

The scientific literature during the last ten years has seen an exponential increase in the number of reports claiming links for genetic polymorphisms with a variety of medical diseases, particularly chronic immune and inflammatory conditions. Recently, periodontal research has contributed to this growth area. This new research has coincided with an increased understanding of the genome which, in turn, has permitted the functional interrelationships of gene products with each other and with environmental agents to be understood. As a result of this knowledge explosion, it is evident that there is a genetic basis for most diseases, including periodontitis. This realization has fostered the idea that if we can understand the genetic basis of diseases, genetic tests to assess disease risk and to develop etiology-based treatments will soon be reality. Consequently, there has been great interest in identifying allelic variants of genes that can be used to assess disease risk for periodontal diseases. Reports of genetic polymorphisms associated with periodontal disease are increasing, but the limitations of such studies are not widely appreciated. While there have been dramatic successes in the identification of mutations responsible for rare genetic conditions, few genetic polymorphisms reported for complex genetic diseases have been demonstrated to be clinically valid, and fewer have been shown to have clinical utility. Although geneticists warn clinicians on the over-enthusiastic use and interpretation of their studies, there continues to be a disparity between the geneticists and the clinicians in the emphasis placed on genes and genetic polymorphism associations. This review critically reviews genetic associations claimed for periodontal disease. It reveals that, despite major advances in the awareness of genetic risk factors for periodontal disease (with the exception of periodontitis associated with certain monogenetic conditions), we are still some way from determining the genetic basis of both aggressive and chronic periodontitis. We have, however, gained considerable insight into the hereditary pattern for aggressive periodontitis. Related to our understanding that it is autosomal-dominant with reduced penetrance comes a major clinically relevant insight into the risk assessment and screening for this disease, in that we appreciate that parents, offspring, and siblings of patients affected with aggressive periodontitis have a 50% risk of this disease also. Nevertheless, we must exercise caution and proper scientific method in the pursuit of clinically valid and useful genetic diagnostic tests for chronic and aggressive periodontitis. We must plan our research using plausible biological arguments and carefully avoid the numerous bias and misinterpretation pitfalls inherent in researching genetic associations with disease.

Generalized T2 test for genome association studies

Recent progress in the development of single-nucleotide polymorphism (SNP) maps within genes and across the genome provides a valuable tool for fine-mapping and has led to the suggestion of genomewide association studies to search for susceptibility loci for complex traits. Test statistics for genome association studies that consider a single marker at a time, ignoring the linkage disequilibrium between markers, are inefficient. In this study, we present a generalized T2 statistic for association studies of complex traits, which can utilize multiple SNP markers simultaneously and considers the effects of multiple disease-susceptibility loci. This generalized T2 statistic is a corollary to that originally developed for multivariate analysis and has a close relationship to discriminant analysis and common measure of genetic distance. We evaluate the power of the generalized T2 statistic and show that power to be greater than or equal to those of the traditional chi2 test of association and a similar haplotype-test statistic. Finally, examples are given to evaluate the performance of the proposed T2 statistic for association studies using simulated and real data.