東華大學圖書館 |

Language: English

Help

回圖書館首頁

手機版館藏查詢

Back

Switch To: Labeled | MARC Mode | ISBD

Non-parametric estimation of distrib...

Wang, Yuanjia.

Linked to FindBook

Google Book

Amazon

博客來

Non-parametric estimation of distribution functions from kin-cohort data.

Record Type:	Electronic resources : Monograph/item
Title/Author:	Non-parametric estimation of distribution functions from kin-cohort data./
Author:	Wang, Yuanjia.
Description:	81 p.
Notes:	Source: Dissertation Abstracts International, Volume: 65-10, Section: B, page: 5222.
Contained By:	Dissertation Abstracts International65-10B.
Subject:	Statistics. -
Online resource:	http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=3151289
ISBN:	0496112414

Non-parametric estimation of distribution functions from kin-cohort data.
Wang, Yuanjia.

Non-parametric estimation of distribution functions from kin-cohort data. - 81 p.

Source: Dissertation Abstracts International, Volume: 65-10, Section: B, page: 5222.

Thesis (Ph.D.)--Columbia University, 2005.

Investigators are often interested in estimating the distribution of gene-related phenotypes as a function of genotypes. When all the genotypes are fairly common, a natural design to estimate the genotype-specific distributions would be random sampling. With a random sample, the conditional distribution of the phenotype given the genotype can be examined directly. However, when there are rare genetic variants, it can be difficult to obtain a sufficient number of carriers to enable accurate estimation of the distribution of the phenotype among carriers of the rare variant. When a rare genotype is indicated by, for example, an extreme value of the phenotype, one may use the presence of the extreme value in developing a case-control sampling plan: the sampling plan would involve a sample of individuals with the extreme value and a sample of individuals without the extreme value. However, while a case-control design could enrich the study sample for rare genetic variants, such a case-control study does not allow for valid estimation of marginal distributions.

ISBN: 0496112414Subjects--Topical Terms:

517247
Statistics.

Non-parametric estimation of distribution functions from kin-cohort data.
LDR:04412nmm 2200301 4500 001 1817489
005 20060814143859.5
008 130610s2005 eng d
020 $a 0496112414
035 $a (UnM)AAI3151289
035 $a AAI3151289
040 $a UnM $c UnM
100 1 $a Wang, Yuanjia. $3 1906842
245 1 0 $a Non-parametric estimation of distribution functions from kin-cohort data.
300 $a 81 p.
500 $a Source: Dissertation Abstracts International, Volume: 65-10, Section: B, page: 5222.
500 $a Adviser: Daniel Rabinowitz.
502 $a Thesis (Ph.D.)--Columbia University, 2005.
520 $a Investigators are often interested in estimating the distribution of gene-related phenotypes as a function of genotypes. When all the genotypes are fairly common, a natural design to estimate the genotype-specific distributions would be random sampling. With a random sample, the conditional distribution of the phenotype given the genotype can be examined directly. However, when there are rare genetic variants, it can be difficult to obtain a sufficient number of carriers to enable accurate estimation of the distribution of the phenotype among carriers of the rare variant. When a rare genotype is indicated by, for example, an extreme value of the phenotype, one may use the presence of the extreme value in developing a case-control sampling plan: the sampling plan would involve a sample of individuals with the extreme value and a sample of individuals without the extreme value. However, while a case-control design could enrich the study sample for rare genetic variants, such a case-control study does not allow for valid estimation of marginal distributions.
520 $a The kin-cohort study design is another approach that may be used when presence of the rare genotype is indicated by extreme values of the phenotype or by any distinguishable range of phenotype values. Like case-control designs, the kin-cohort design results in samples enriched for the rare genetic variants. Unlike case-control design, however, all aspects of genotype-specific distributions generally are identifiable from data from kin-cohort sampling designs. In such a design, genotype and phenotype status are obtained from a group of subjects, the probands, sampled from individuals with phenotypes in the range. The probands are genotyped and phenotype status is then obtained from relatives of the probands. Phenotype status may be obtained either by interviewing the probands or by interviewing the relatives themselves. In the kin-cohort design relatives of probands are generally not genotyped. By examining the relatives of the probands researchers can enrich a sample with carriers. This is because the probands are enriched for the rare variants carriers and so that their relatives are likely to be carriers too. Kin-cohort design trades issues related to incomplete genotype data in family members for greater efficiency than may be possible with a cohort design.
520 $a This thesis presents statistical methods for analyzing data from kin-cohort studies. A class of non-parametric estimators of genotype-specific distributions of a phenotype is developed, and the optimal member of the class is derived. It is shown that the optimal estimator in the class lies in the tangent space of the non-parametric model, and is therefore efficient among the class of regular unbiased estimators. A two-step approach to approximating the optimal estimator is described. The approximate estimator is shown to be asymptotically equivalent to the optimal estimator and therefore asymptotically efficient. Simulation studies are used to examine numerical properties of the method, including the relative efficiency of the optimal estimator and the approximate optimal estimator compared to simple estimators. Variance estimators are also examined. The method is illustrated through an analysis of data on the Parkin gene mutation and age-at-onset of Parkinson's disease. The data used in the illustrative data analysis is from the study "Familial Aggregation of Early- and Late-Onset Parkinson's Disease" (Marder et al. 2003(a)).
590 $a School code: 0054.
650 4 $a Statistics. $3 517247
650 4 $a Biology, Genetics. $3 1017730
690 $a 0463
690 $a 0369
710 2 0 $a Columbia University. $3 571054
773 0 $t Dissertation Abstracts International $g 65-10B.
790 1 0 $a Rabinowitz, Daniel, $e advisor
790 $a 0054
791 $a Ph.D.
792 $a 2005
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=3151289