對RILXP1族群進行數量性狀基因座定位的統計方法研究

Abstract

回交和F2族群是數量性狀基因座定位(QTL mapping)中最普遍流行的試驗族群。近年來，部分高階族群、重組自交系(RILs)等因為其族群結構與F2族群有所不同，提供了改進QTL mapping的可能性，而漸漸受到重視。在本文中，我們以RILXP1族群作為試驗族群，利用區間定位法及多重區間定位法進行QTL mapping。在估計上，區間定位法的統計模式是一個混合常態模式，其混合比例可由待測數量性狀基因座在已知兩翼標識的基因型下的條件機率決定。當以RILXP1族群為試驗族群進行QTL mapping時，過去所提出的條件機率計算公式並不正確，我們提出正確的計算方式，並以分析和模擬的方式比較兩種方法在不同的遺傳率、聯鎖強度、交感作用等等條件下對數量性狀基因座的偵測能力及估計上的差異。

Backcross and F2 populations are the most popular and common experimental populations for QTL mapping. Recently, some advanced populations such as the recombinant inbred population and recombinant inbred lines (RILs) are of growing interest since these populations have different population structures from that of F2 population and hence providing the opportunity to achive better QTL mapping resolution. In this thesis, the RILXP1 population is considered for QTL mapping, and interval mapping and multiple interval mapping are used to detect QTLs. In estimation, the statistical model for interval mapping procedure is usually a normal mixture model, and its mixing proportions can be determined by the conditional probabilities of the QTL genotypes given the flanking marker genotypes. As the formulas suggested for computing the conditional probabilities by the traditional approach were not correct, we proposed the correct formulas to compute conditional probabilities so that QTL mapping can be improved. Simulations are performed to investigate the difference in the estimation and mapping power between the methods using the two different formulas under different linkage strengths, epistasis and relative effect sizes.