東華大學圖書館 |

語系: 繁體中文

說明(常見問題)

回圖書館首頁

手機版館藏查詢

登入

回首頁

切換: 標籤 | MARC模式 | ISBD

Genetic association analyses of time...

Boston University.

FindBook

Google Book

Amazon

博客來

Genetic association analyses of time-to-event data: Selection bias and imputation with examples from the Framingham Heart Study.

紀錄類型:	書目-語言資料,印刷品 : Monograph/item
正題名/作者:	Genetic association analyses of time-to-event data: Selection bias and imputation with examples from the Framingham Heart Study./
作者:	Yin, Xiaoyan.
面頁冊數:	169 p.
附註:	Adviser: Martin G. Larson.
Contained By:	Dissertation Abstracts International69-08B.
標題:	Biology, Biostatistics. -
電子資源:	http://pqdd.sinica.edu.tw/twdaoeng/servlet/advanced?query=3323155
ISBN:	9780549749585

Genetic association analyses of time-to-event data: Selection bias and imputation with examples from the Framingham Heart Study.
Yin, Xiaoyan.

Genetic association analyses of time-to-event data: Selection bias and imputation with examples from the Framingham Heart Study. - 169 p.

Adviser: Martin G. Larson.

Thesis (Ph.D.)--Boston University, 2009.

In genetic association studies, especially in longitudinal cohort studies, subjects may exist who have phenotypic data but who have no genetic data. One analysis option is to study those who have complete data on phenotypes and genotypes (complete case analysis), but this approach has lower precision and may introduce bias if those missing genotype data are not representative of the full cohort. For example, the Offspring Cohort of the Framingham Heart Study was genotyped only partially because blood for DNA samples was collected about 30 years after enrollment. Participants who died before sample collection could not contribute genetic data. Thus, results for genes associated with survival may be misrepresented in a complete case analysis. In this thesis, I examine methods for detecting mortality-related genetic variants. First, I explore the use of Hardy-Weinberg equilibrium and Mantel's test of trend for detecting such variants. Mantel's trend test for correlation between age group and genotype distribution is practical and has good power. Second, I studied the genetic effect associated with an intermediate event such as time to cardiovascular disease. I used simulations to study the bias caused by analyzing only genotyped individuals. The bias is away from the null hypothesis that there is no association between a genetic variant and the phenotype; its magnitude depends on the true genetic effect size and the proportion of missingness. In general, the bias is 4% to 6% for a typical middle-aged cohort after 30 years of follow up. Third, in the context of Cox's proportional hazard models for time to event, I develop two easily applied approaches for imputing genotype data for individuals who were not genotyped: (i) multiple imputation based on predicted genotype probabilities, and (ii) an expectation-maximization (EM)-like iterative approach using weights. Both methods reduce bias and improve estimation precision compared with complete case analysis. The methods can be implemented using routine statistical packages such as Statistical Analysis System (SAS). In summary, this thesis expands the collection of methods for studying mortality related genetic variants and methods for studying genotype effect on intermediate events with incompletely observed genetic data.

ISBN: 9780549749585Subjects--Topical Terms:

1018416
Biology, Biostatistics.

Genetic association analyses of time-to-event data: Selection bias and imputation with examples from the Framingham Heart Study.
LDR:03214nam 2200265 a 45 001 856707
005 20100709
008 100709s2009 ||||||||||||||||| ||eng d
020 $a 9780549749585
035 $a (UMI)AAI3323155
035 $a AAI3323155
040 $a UMI $c UMI
100 1 $a Yin, Xiaoyan. $3 1023557
245 1 0 $a Genetic association analyses of time-to-event data: Selection bias and imputation with examples from the Framingham Heart Study.
300 $a 169 p.
500 $a Adviser: Martin G. Larson.
500 $a Source: Dissertation Abstracts International, Volume: 69-08, Section: B, page: 4510.
502 $a Thesis (Ph.D.)--Boston University, 2009.
520 $a In genetic association studies, especially in longitudinal cohort studies, subjects may exist who have phenotypic data but who have no genetic data. One analysis option is to study those who have complete data on phenotypes and genotypes (complete case analysis), but this approach has lower precision and may introduce bias if those missing genotype data are not representative of the full cohort. For example, the Offspring Cohort of the Framingham Heart Study was genotyped only partially because blood for DNA samples was collected about 30 years after enrollment. Participants who died before sample collection could not contribute genetic data. Thus, results for genes associated with survival may be misrepresented in a complete case analysis. In this thesis, I examine methods for detecting mortality-related genetic variants. First, I explore the use of Hardy-Weinberg equilibrium and Mantel's test of trend for detecting such variants. Mantel's trend test for correlation between age group and genotype distribution is practical and has good power. Second, I studied the genetic effect associated with an intermediate event such as time to cardiovascular disease. I used simulations to study the bias caused by analyzing only genotyped individuals. The bias is away from the null hypothesis that there is no association between a genetic variant and the phenotype; its magnitude depends on the true genetic effect size and the proportion of missingness. In general, the bias is 4% to 6% for a typical middle-aged cohort after 30 years of follow up. Third, in the context of Cox's proportional hazard models for time to event, I develop two easily applied approaches for imputing genotype data for individuals who were not genotyped: (i) multiple imputation based on predicted genotype probabilities, and (ii) an expectation-maximization (EM)-like iterative approach using weights. Both methods reduce bias and improve estimation precision compared with complete case analysis. The methods can be implemented using routine statistical packages such as Statistical Analysis System (SAS). In summary, this thesis expands the collection of methods for studying mortality related genetic variants and methods for studying genotype effect on intermediate events with incompletely observed genetic data.
590 $a School code: 0017.
650 4 $a Biology, Biostatistics. $3 1018416
690 $a 0308
710 2 $a Boston University. $3 1017454
773 0 $t Dissertation Abstracts International $g 69-08B.
790 $a 0017
790 1 0 $a Larson, Martin G., $e advisor
791 $a Ph.D.
792 $a 2009
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoeng/servlet/advanced?query=3323155