131 research outputs found
Statistical Modeling and Testing for Joint Association in Genome-Wide Association Studies
No full textUniversity of Minnesota Ph.D. dissertation. July 2015. Major: Biostatistics. Advisor: Saonli Basu. 1 computer file (PDF); ix, 139 pages.Most common human diseases are complex genetic traits, with multiple genetic and environmental components contributing to the disease susceptibility. Genome-wide Association Studies (GWASs) offer a powerful approach to identify the genetic variants (single nucleotide polymorphisms or SNPs) that modulate the susceptibility to these complex diseases. GWASs have identified hundreds of SNPs associated with such diseases, but these SNPs appear to explain very little of the genetic risk. This dissertation aims at investigating several alternative hypotheses for explaining the disease risk and develop statistical techniques to improve the power to detect SNPs influencing such diseases. A Bayesian dimension reduction model is developed to study the joint effect of a group of SNPs on the disease status for unrelated individuals. Modeling the joint effects of multiple SNPs can help in the detection of SNPs that jointly have significant risk effects but individually make only a small contribution. Thus, our method based on the proposed dimension reduction model, Bayesian partitioning model (BPM), may have improved power over multiple single-SNP association analysis when testing the association of multiple SNPs with a single binary trait. Similarly, joint analysis of multiple disease-related traits may also improve detection of SNPs associated with a disease. GWASs often collect data on multiple disease-related traits. These traits may share a common set of SNPs influencing them and a joint analysis of these traits may improve the power to detect these SNPs which may provide a better understanding of the underlying disease mechanism. Multivariate analysis of variance (MANOVA) can perform such an association analysis at a GWAS level. The behavior of MANOVA is investigated, both theoretically and using simulations, and the conditions where MANOVA loses power are derived. Based on these findings, a novel unified score-based association test (USAT) is proposed that adaptively uses the data to optimize power to detect association of a single SNP with multiple quantitative phenotypes/traits. This test and other such multivariate tests are based on the assumption of random sampling, and may suffer from severely inflated type I error in case of selected sampling. This motivated us to explore scenarios in which popular methods would fail to provide valid tests of the null hypothesis of no association of a single SNP with multiple traits within the framework of a case-control study. Two alternative hypothesis testing approaches (one based on maximum p-value and the other based on propensity score) are proposed for such scenarios.Ray, Debashree. (2015). Statistical Modeling and Testing for Joint Association in Genome-Wide Association Studies. Retrieved from the University Digital Conservancy, https://hdl.handle.net/11299/190503
STATISTICAL METHODS FOR INTEGRATING DISPARATE DATA SOURCES
No full textMy thesis is about developing statistical methods by integrating disparate data sources with real data applications, and identifying gene-environment interactions (G x E) in more extensive studies using existing analytical methods. We propose a general and novel statistical framework for combining information on multivariate
regression parameters across multiple different studies which have varying level of covariate information . We illustrate the method using real data for developing a breast cancer risk prediction model. We propose a generalized method of moments (GMM) approach for analyzing two-phase studies where we take into account the dependent structure of the datasets across the two-phases. We illustrate the method using real data on Wilm's tumor, a common type of kidney cancer in children. We analyze the largest gene by smoking interaction study for pancreatic ductal adenocarcinoma risk conducted to date using existing statistical methods
Validation of reference genes for gene expression analysis in olive (Olea europaea) mesocarp tissue by quantitative real-time RT-PCR
No full textEffect of non-normality and low count variants on cross-phenotype association tests in GWAS
No full text
Integrative and Transfer Learning Methods for Disparate Data with Applications in Single-cell Genomics and Statistical Genetics
No full textIn recent research of genomics and genetics, we face the challenges how to perform data analysis better by incorporating more information either internally from dataset itself or externally from other studies. My thesis will discuss about three approaches to addressing this problem.
The first part reviews existing benchmark works and introduces a new method, \texttt{mixhvg}, for selecting highly variable genes in single-cell RNA-sequencing. This process is vital due to the intrinsic characteristics of single-cell RNA-sequencing. Our work not only fills the gap in comprehensive benchmarks for selecting optimal methods but also proposes \texttt{mixhvg}, a hybrid approach that enhances performance robustly.
The second part builds on the first, examining the effects of highly variable gene selection on downstream analysis, specifically visualization. We find that selected genes depend on data, where local structures benefit from corresponding local gene selections. To illustrate local patterns more effectively, we propose an adaptive method, \texttt{SAVIS}, which integrates with \texttt{mixhvg} to further improve outcomes.
The third part discusses heterogeneous transfer learning for disparate datasets with unmatched feature sets. We address this by exploring transfer learning between two data types to construct high-dimensional generalized linear models. Here, the primary dataset has a smaller sample size but a comprehensive set of variables of interest, while the external dataset is larger but feature-limited. Our proposed HTL-GMM method utilizes the generalized method of moments (GMM) to enhance both prediction accuracy and post-selection inference effectively.
Overall, this thesis focuses on integrative and transfer learning methods applicable to single-cell genomics and genetic data, aiming to advance analytical capabilities in these fields
Integrative and Transfer Learning Methods for Disparate Data with Applications in Single-cell Genomics and Statistical Genetics
No full textIn recent research of genomics and genetics, we face the challenges how to perform data analysis better by incorporating more information either internally from dataset itself or externally from other studies. My thesis will discuss about three approaches to addressing this problem.
The first part reviews existing benchmark works and introduces a new method, \texttt{mixhvg}, for selecting highly variable genes in single-cell RNA-sequencing. This process is vital due to the intrinsic characteristics of single-cell RNA-sequencing. Our work not only fills the gap in comprehensive benchmarks for selecting optimal methods but also proposes \texttt{mixhvg}, a hybrid approach that enhances performance robustly.
The second part builds on the first, examining the effects of highly variable gene selection on downstream analysis, specifically visualization. We find that selected genes depend on data, where local structures benefit from corresponding local gene selections. To illustrate local patterns more effectively, we propose an adaptive method, \texttt{SAVIS}, which integrates with \texttt{mixhvg} to further improve outcomes.
The third part discusses heterogeneous transfer learning for disparate datasets with unmatched feature sets. We address this by exploring transfer learning between two data types to construct high-dimensional generalized linear models. Here, the primary dataset has a smaller sample size but a comprehensive set of variables of interest, while the external dataset is larger but feature-limited. Our proposed HTL-GMM method utilizes the generalized method of moments (GMM) to enhance both prediction accuracy and post-selection inference effectively.
Overall, this thesis focuses on integrative and transfer learning methods applicable to single-cell genomics and genetic data, aiming to advance analytical capabilities in these fields
sj-docx-1-cpc-10.1177_10556656221135926 - Supplemental material for Damaging Mutations in <b><i>AFDN</i></b> Contribute to Risk of Nonsyndromic Cleft Lip With or Without Cleft Palate
No full textSupplemental material, sj-docx-1-cpc-10.1177_10556656221135926 for Damaging Mutations in AFDN Contribute
to Risk of Nonsyndromic Cleft Lip With or Without Cleft Palate by Waheed Awotoye, Peter A Mossey, Jacqueline B Hetmanski, Lord J J Gowans, Mekonen A Eshete and
Wasiu L Adeyemo, Azeez Alade, Erliang Zeng, Olawale Adamson, Olutayo James,
Azeez Fashina, Modupe O Ogunlewe,
Thirona Naicker, Chinyere Adeleke, Tamara Busch,
Mary Li, Aline Petrin, Abimbola Oladayo, Sami Kayali, Joy Olotu, Veronica Sule, Mohaned Hassan, John Pape, Emmanuel T Aladenika, Peter Donkor, Fareed K N Arthur, Solomon Obiri-Yeboah, Daniel K Sabbah, Pius Agbenorku, Debashree Ray,
Gyikua Plange-Rhule, Alexander Acheampong Oti,
Daniah Albokhari, Nara Sobreira, Martine Dunnwald, Terri H Beaty, Margaret Taub, Mary L Marazita,
Adebowale A Adeyemo, Jeffrey C Murray, Azeez Butali in The Cleft Palate-Craniofacial Journal</p
Maternal and Parent-of-Origin Gene–Environment Effects on the Etiology of Orofacial Clefting
No full textBackground/Objectives: We investigated maternal and parent-of-origin (PoO) gene-environment interaction effects on the risk of nonsyndromic orofacial clefts for two maternal environmental factors: periconceptional smoking and folic acid supplementation. Methods: Genome-wide single nucleotide polymorphisms (SNPs) genotypes and TopMed-imputed genotypes were obtained for case-parent triads from the EUROCRAN and ITALCLEFT studies. Candidate regions were selected around target SNPs from a previous genome-wide association study, resulting in 12 (726 SNPs) and 11 regions (730 SNPs) for maternal and PoO effects, respectively. Log-linear models were used to analyze 404 case-parent triads and 40 case-parent dyads. p-values were combined across regions. Results: None of the interactions reached statistical significance after correction for the number of regions tested. Nominally significant (pooled p-values < 0.05) interactions pointed to regions in or close to genes LRRC7 (maternal gene-folate interaction), NCKAP5 (PoO-smoking interaction), and IFT43 and GPATCH2L (PoO-folate interaction). Conclusions: Our results suggested that the genetic effects in or around these genes were heightened under periconceptional exposure to tobacco or no folic acid supplementation. The involvement of these genes in orofacial cleft development, in conjunction with environmental exposures, should be further studied.</p
- …
