2026-09-19T22:23:35Zhttps://digital.lib.washington.edu/server/oai/request

oai:digital.lib.washington.edu:1773/260042026-02-16T02:29:48Zcom_1773_4888col_1773_4900

Edlefsen, Paul Kee, Jia Jin 2014-10-13T16:50:39Z 2014-10-13T16:50:39Z 2014-10-13 2014 Kee_washington_0250O_13682.pdf http://hdl.handle.net/1773/26004 Thesis (Master's)--University of Washington, 2014 Amino acid substitution matrices are commonly used for sequence alignment, phylogenetic inference and sequence comparison. Empirical organism-specific substitution matrices constructed using only sequence data from a particular organism are thought to lead to more accurate analyses. In HIV research, the standard substitution matrices are the between- and within-host matrices estimated using HIV sequences introduced by Nickle et al. (2007). This thesis focuses on constructing more granular HIV-specific matrices (clade-specific matrices and gene-specific matrices) and comparing the matrices in a way that accounts for error in matrix estimation. Using standard errors of parameter estimates predicted from a two-part linear model, the analyses indicate statistically significant difference between HIV clade B and HIV clade C matrices as well as between HIV Env gene and HIV Gag gene matrices upon performing Bonferroni-corrected comparisons of 189 estimates of amino acid exchangeability parameters. application/pdf en_US Copyright is held by the individual authors. Biostatistics Bioinformatics biostatistics Estimation and Comparison of HIV-Specific Substitution Matrices Thesis Open Access open.access