使用模擬資料來比較選取標籤單體核苷酸多樣性、分單體型區塊與相關性檢定所用不同方法之組合

Abstract

使用病例-對照研究方法的母群體相關性檢定對於偵測人類一般疾病的治病基因有顯著功效，我們感興趣的是比較不同選取標籤單體核□酸多態性的方法與不同相關性檢定方法在不同樣本下的檢定力。 我們使用四種選取標籤單體核□酸多態性方法，其中三種方法是基於單體型多樣性，一種是基於兩兩連鎖不平衡的觀點；並選用四種相關性檢定的方法，其中三種是基於多重單體核□酸多態性，一種是基於單一單體核□酸多態性的觀點，此外，分單體型區塊的方法也是我們考慮的一項因素。 我們從一組名為「第十五次基因分析專題討論模擬資料」中選出兩段區域來估計每種方法的檢定力與型一錯誤。 相關性檢定中，使用多重單體核□酸多態性的方法會比使用單一單體核□酸多態性的方法具有較高的檢定力，且在做標籤單體核□酸多態性時選用疾病樣本會比選用非疾病樣本有較高的檢定力。 樣本大小是影響相關性檢定的檢定力之重要因素，但對於做標籤單體核□酸多態性時並不會有顯著的影響，最後我們建議了幾種在做單體核□酸多態性時較好的方法組合。

Population association studies with case-control designs are powerful to detect the genetic variations responsible for human common diseases. We were interested in how the tag SNP selection methods with association tests and samples used for tag SNP discovery would have on power. We used four methods for choosing tag SNPs: three based on haplotype diversity, one based on pair-wise linkage disequilibrium (LD) and four methods to detect association: three based on multiple-SNP test, one based on single-SNP test. Besides, haplotype blocking is an important factor we considered. In two regions from the Genetic Analysis Workshop 15 simulated data, we estimated the power and type I error at each match. The multiple-SNP test is more power than single-SNP test. In most situations, the case sample used for tag SNP selection is more power than the control sample. Association sample sizes are evident reason to effect power but tag SNP selection sample sizes are not significant to power. At the end, we advised some combinations of methods for SNP analysis.