Skip to Main Content
 

Global Search Box

 
 
 
 

Files

File List


osu1164815591.pdf (951.45 KB)

ETD Abstract Container

Abstract Header

Likelihood-based procedures for obtaining confidence intervals of disease Loci with general pedigree data

Wan, Shuyan
http://rave.ohiolink.edu/etdc/view?acc_num=osu1164815591

Abstract Details

2006, Doctor of Philosophy, Ohio State University, Statistics.
Mapping of complex traits has become the main challenge in the field of linkage analysis. To localize a disease gene in a whole-genome study, the traditional testing of linkage signals turns out to be lack of accurate assessment of statistical significance. In this dissertation, we propose to test every position on the chromosome for its possibility of being a disease locus and construct a confidence set for the disease locus by including those positions that are not rejected. Three test statistics are proposed to perform the test, including a LOD and two generalized likelihood ratio test statistics with or without model averaging (GLRT/MA and GLRT). Based on different test statistics, several approaches are developed to estimate the null distribution. Specifically, based on LOD, an integrated procedure is formed by an adaptive procedure and an importance sampling procedure. Asymptotic approaches based on GLRT and GLRT/MA are also proposed as alternatives that are much more efficient computationally but rely on the validity of the limiting distribution of approximation. An importance sampling procedure is also developed based on GLRT (GLRT/IS). Simulation studies show that the integrated procedure tends to give longer confidence intervals with right coverage probability. Both the GLRT asymptotic and the GLRT/IS approaches give actual coverage probability around the nominal one when there are moderate to strong linkage signals. When the signals are weak, the GLRT asymptotic approach tends to undercover the disease locus while the GLRT/IS has the correct coverage. Compared with the 1-LOD support interval, all our methods capture the disease loci to confidence intervals with closer to nominal coverage probability when there are weak to moderate linkage signals. However, when there is strong linkage information, the performance of the 1-LOD method is as good as that of the GLRT/IS. Finally, we apply all the methods to three sets of rheumatoid arthritis data to obtain confidence intervals of the putative disease loci. Both asymptotic methods and the integrated procedure recover the well recognized susceptibility locus, HLA*DRB1, to 99% confidence intervals of around 5-20 cM in length while the 3-LOD method either misses it or gets a longer confidence set.
Shili Lin (Advisor)
Joseph Verducci (Other)
Mei-Ling Lee (Other)
Statistics
LOD; Importance Sampling; Linkage Analysis; CSI

Recommended Citations

Citations

  • Wan, S. (2006). Likelihood-based procedures for obtaining confidence intervals of disease Loci with general pedigree data [Doctoral dissertation, Ohio State University]. OhioLINK Electronic Theses and Dissertations Center. http://rave.ohiolink.edu/etdc/view?acc_num=osu1164815591

    APA Style (7th edition)

  • Wan, Shuyan. Likelihood-based procedures for obtaining confidence intervals of disease Loci with general pedigree data. 2006. Ohio State University, Doctoral dissertation. OhioLINK Electronic Theses and Dissertations Center, http://rave.ohiolink.edu/etdc/view?acc_num=osu1164815591.

    MLA Style (8th edition)

  • Wan, Shuyan. "Likelihood-based procedures for obtaining confidence intervals of disease Loci with general pedigree data." Doctoral dissertation, Ohio State University, 2006. http://rave.ohiolink.edu/etdc/view?acc_num=osu1164815591

    Chicago Manual of Style (17th edition)

osu1164815591
620
© 2006, all rights reserved.
This open access ETD is published by The Ohio State University and OhioLINK.