Novel Statistical Methods in Quantitative Genetics : Modeling Genetic Variance for Quantitative Trait Loci Mapping and Genomic Evaluation

This thesis develops and evaluates statistical methods for different types of genetic analyses, including quantitative trait loci (QTL) analysis, genome-wide association study (GWAS), and genomic evaluation. The main contribution of the thesis is to provide novel insights in modeling genetic variance, especially via random effects models. In variance component QTL analysis, a full likelihood model accounting for uncertainty in the identity-by-descent (IBD) matrix was developed. It was found to be able to correctly adjust the bias in genetic variance component estimation and gain power in QTL mapping in terms of precision.  Double hierarchical generalized linear models, and a non-iterative simplified version, were implemented and applied to fit data of an entire genome. These whole genome models were shown to have good performance in both QTL mapping and genomic prediction. A re-analysis of a publicly available GWAS data set identified significant loci in Arabidopsis that control phenotypic variance instead of mean, which validated the idea of variance-controlling genes.  The works in the thesis are accompanied by R packages available online, including a general statistical tool for fitting random effects models (hglm), an efficient generalized ridge regression for high-dimensional data (bigRR), a double-layer mixed model for genomic data analysis (iQTL), a stochastic IBD matrix calculator (MCIBD), a computational interface for QTL mapping (qtl.outbred), and a GWAS analysis tool for mapping variance-controlling loci (vGWAS).

A control chart based on sample ranges for monitoring the covariance matrix of the multivariate processes

Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq)

A new chart based on sample variances for monitoring the covariance matrix of multivariate processes

Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq)

A new chart for monitoring the covariance matrix of bivariate processes

In this article we consider a control chart based on the sample variances of two quality characteristics. The points plotted on the chart correspond to the maximum value of these two statistics. The main reason to consider the proposed chart instead of the generalized variance |S| chart is its better diagnostic feature, that is, with the new chart it is easier to relate an out-of-control signal to the variables whose parameters have moved away from their in-control values. We study the control chart efficiency considering different shifts in the covariance matrix. In this way, we obtain the average run length (ARL) that measures the effectiveness of a control chart in detecting process shifts. The proposed chart always detects process disturbances faster than the generalized variance |S| chart. The same is observed when the size of the samples is variable, except in a few cases in which the size of the samples switches between small size and very large size.

The Synthetic Control Chart Based on Two Sample Variances for Monitoring the Covariance Matrix

Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq)

Monitoring the covariance matrix with the VMAX chart

In this article, we propose a new statistic to control the covariance matrix of bivariate processes. This new statistic is based on the sample vat-lances of the two quality characteristics, shortly VMAX statistic. The points plotted on the chart correspond to the maximum of the values of these two variances. The reasons to consider the VMAX statistic instead of the generalized variance vertical bar S vertical bar are faster detection of process changes and better diagnostic feature, that is, with the VMAX statistic It is easier to identify the out-of-control variable.

A single statistic for monitoring the covariance matrix of bivariate processes

Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq)

Estimates of genetic parameters and eigenvector indices for milk production of Holstein cows

The objectives of the present study were to estimate genetic parameters of monthly test-day milk yield (TDMY) of the first lactation of Brazilian Holstein cows using random regression (RR), and to compare the genetic gains for milk production and persistency, derived from RR models, using eigenvector indices and selection indices that did not consider eigenvectors. The data set contained monthly TDMY of 3,543 first lactations of Brazilian Holstein cows calving between 1994 and 2011. The RR model included the fixed effect of the contemporary group (herd-month-year of test days), the covariate calving age (linear and quadratic effects), and a fourth-order regression on Legendre orthogonal polynomials of days in milk (DIM) to model the population-based mean curve. Additive genetic and nongenetic animal effects were fit as RR with 4 classes of residual variance random effect. Eigenvector indices based on the additive genetic RR covariance matrix were used to evaluate the genetic gains of milk yield and persistency compared with the traditional selection index (selection index based on breeding values of milk yield until 305 DIM). The heritability estimates for monthly TDMY ranged from 0.12 ± 0.04 to 0.31 ± 0.04. The estimates of additive genetic and nongenetic animal effects correlation were close to 1 at adjacent monthly TDMY, with a tendency to diminish as the time between DIM classes increased. The first eigenvector was related to the increase of the genetic response of the milk yield and the second eigenvector was related to the increase of the genetic gains of the persistency but it contributed to decrease the genetic gains for total milk yield. Therefore, using this eigenvector to improve persistency will not contribute to change the shape of genetic curve pattern. If the breeding goal is to improve milk production and persistency, complete sequential eigenvector indices (selection indices composite with all eigenvectors) could be used with higher economic values for persistency. However, if the breeding goal is to improve only milk yield, the traditional selection index is indicated. © 2013 American Dairy Science Association.

A Single Statistic for Monitoring the Covariance Matrix of Bivariate Processes

In this article, we present a new control chart for monitoring the covariance matrix in a bivariate process. In this method, n observations of the two variables were considered as if they came from a single variable (as a sample of 2n observations), and a sample variance was calculated. This statistic was used to build a new control chart specifically as a VMIX chart. The performance of the new control chart was compared with its main competitors: the generalized sampled variance chart, the likelihood ratio test, Nagao's test, probability integral transformation (v(t)), and the recently proposed VMAX chart. Among these statistics, only the VMAX chart was competitive with the VMIX chart. For shifts in both variances, the VMIX chart outperformed VMAX; however, VMAX showed better performance for large shifts (higher than 10%) in one variance.

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.

Dissecting the complex genetic basis of mate choice

The genetic analysis of mate choice is fraught with difficulties. Males produce complex signals and displays that can consist of a combination of acoustic, visual, chemical and behavioural phenotypes. Furthermore, female preferences for these male traits are notoriously difficult to quantify. During mate choice, genes not only affect the phenotypes of the individual they are in, but can influence the expression of traits in other individuals. How can genetic analyses be conducted to encompass this complexity? Tighter integration of classical quantitative genetic approaches with modern genomic technologies promises to advance our understanding of the complex genetic basis of mate choice.

Genetic Polymorphisms of Matrix Metalloproteinases: Susceptibility and Prognostic Implications for Prostate Cancer

Purpose: Prostate cancer is the most common tumor in males in Brazil. Single nucleotide polymorphisms have been demonstrated to exist in the promoter regions of matrix metalloproteinase genes and they are associated with the development and progression of some cancers. We investigated the correlation between MMP1, 2, 7 and 9 polymorphisms with susceptibility to prostate cancer, and classic prognostic parameters of prostate cancer. Materials and Methods: Genomic DNA was extracted using conventional protocols. The DNA sequence containing the polymorphic site was amplified by realtime polymerase chain reaction using TaqMan (R) fluorescent probes. Results: For the MMP1 gene the polymorphic allele was more common in the control group than in the prostate cancer group (p <0.001). For the MMP9 gene the incidence of the polymorphic homozygote genotype was higher in the prostate cancer group (p <0.001). For higher stage tumors (pT3) a polymorphic allele in the MMP2 gene was more common (p = 0.026). When considering Gleason score, the polymorphic homozygote genotype of MMP9 was more common in Gleason 6 or less tumors (p = 0.003), while a polymorphic allele in the MMP2 gene was more common in Gleason 7 or greater tumors (p = 0.042). Conclusions: MMP1 and MMP2 may protect against prostate cancer development and MMP9 may be related to higher risk. In contrast, MMP9 polymorphism was associated with a lower Gleason score and MMP2 polymorphism was associated with nonorgan confined disease.

A genetic polymorphism of matrix metalloproteinase 9 (MMP-9) affects the changes in circulating MMP-9 levels induced by highly active antiretroviral therapy in HIV patients

We examined whether two functional polymorphisms (g.-1562C>T and g.-90(CA)14-24) in the matrix metalloproteinase (MMP)-9 gene or MMP-9 haplotypes affect the circulating levels of pro-MMP-9 and pro-MMP-9/TIMP-1 (tissue inhibitor of metalloproteinase-1) ratios in AIDS patients, and modulate alterations in these biomarkers after highly active antiretroviral therapy (HAART). We studied 82 patients commencing HAART. Higher pro-MMP-9 concentrations and pro-MMP-9/TIMP-1 ratios were found in CT/TT patients compared with CC patients. HAART decreased pro-MMP-9 levels and pro-MMP-9/TIMP-1 ratios in CT/TT patients, it did not modify pro-MMP-9 levels and it increased pro-MMP-9/TIMP-1 ratios in CC patients. The g.-90(CA)14-24 polymorphism, however, produced no significant effects. Moreover, we found no significant differences in HAART-induced changes in plasma pro-MMP-9, TIMP-1 and pro-MMP-9/TIMP-1 ratios when different MMP-9 haplotypes were compared. These findings suggest that the g.-1562C>T polymorphism affects pro-MMP-9 levels in patients with AIDS and modulates the alterations in pro-MMP-9 levels caused by HAART, thus possibly affecting the risk of cardiovascular complications. The Pharmacogenomics Journal (2009) 9, 265-273; doi: 10.1038/tpj.2009.13; published online 21 April 2009

Evolution of additive and nonadditive genetic variance in development time along a cline in Drosophila serrata

Latitudinal clines provide natural systems that may allow the effect of natural selection on the genetic variance to be determined. Ten clinal populations of Drosophila serrata collected from the eastern coast of Australia were used to examine clinal patterns in the trait mean and genetic variance of the life-history trait egg-to-adult development time. Development time significantly lengthened from tropical areas to temperate areas. The additive genetic variance for development time in each population was not associated with latitude but was associated with the population mean development time. Additive genetic variance tended to be larger in populations with more extreme development times and appeared to be consistent with allele frequency change. In contrast, the nonadditive genetic variance was not associated with the population mean but was associated with latitude. Levels of nonadditive genetic variance were greatest in the region of the cline where the gradient in the change in mean was greatest, consistent with Barton's (1999) conjecture that the generation of linkage disequilibrium may become an important component of the genetic variance in systems with a spatially varying optimum.

Jamming detection in GNSS signals using the sample covariance matrix.

We propose a blind method to detect interference in GNSS signals whereby the algorithms do not require knowledge of the interference or channel noise features. A sample covariance matrix is constructed from the received signal and its eigenvalues are computed. The generalized likelihood ratio test (GLRT) and the condition number test (CNT) are developed and compared in the detection of sinusoidal and chirp jamming signals. A computationally-efficient decision threshold was proposed for the CNT.