高三酸甘油酯／低高密度脂蛋白膽固醇之家族遺傳研究(3/3)

Abstract

本年度計畫則是以金山社區的 Triglyceride/HDL-cholesterol 當作一個連 續性性狀，在ApoA1/C3/A4 基因簇集中找 到可能與之相關的控制基因，在基本家族 資料成員，在所有456 個家族，共4175 位 家族成員(包括1436 位founders 及2739 位 non-founders)，家族大小平均為9.16 人(範 圍1~67)，有26.8%為7 人，有16. 9%為6 人，有14.0%為8 人，共2.78 代(範圍1~5 代)，有68.9%為3 代，有22.6%為2 代， 有5.0%為4 代，而真正有基因型資料，約 為20%，而異體基因 (heterozygosis)約有 40%左右。參數性連鎖分析可用在以下幾 種狀況：藉由估計互組事件的數目來評估 基因標識與疾病相關位址的基因距離, 如 果組合率(Recombination fraction ，θ)已 知，可進一步在基因地圖上作排序, 針對常 見疾病找出其持定的基因型式。本研究資 料結果顯示，其最高LOD Score 分別發生 在組合率(θ)不同的狀況下，其估計的full informative meiosis 相等數量分別為 0.1197(θ=0.1)，也即是資料提供的訊息量 等同於0.1197 個full informative meiosis 。 結果進一步顯示其最大LOD Score 是在0 ≦θ≦1 之間，雖然結果並無法達到統計上 有意義的LOD Score ，但了解其θ變化及 不同的家族供lod score 有其必要性。作定 量性狀遺傳分析，有基因標記之下的流 程：先以Merlin 程式作錯誤偵測程式方 式，特別針對是否符合Mendel 遺傳定律， 在每個SNP 簇集檢查，以最大概似值方法 去偵測某些基因型明顯此預期的概似值為 大，並且比較以偵測出前Error 及之後的資 料跑Merlin ，可看到mismatched 資料之後 可得到較高的LOD Score(在迴歸模型中 LOD Score -0.035 增加為0.827 ，在變異數 成份模型中LOD Score 由0.60 (ApoA75)變 成0.86 ，在迴歸模型中ApoA25 LOD Score 也變大，可以增加其檢定效力，對偵測可 能的基因位址有較大成功的機率。

The basic description of family data was following; Individuals: 4175 (1436 founders, 2739 nonfounders), and total families: 456. The family sizes were average: 9.16 (1 to 67), and the distribution patterns of family were 7 (26.8%), 6 (16.9%) and 8 (14.0%). The generations were average: 2.78 (1 to 5). The estimate of residual nonshared environmental variance is 0.324, and the estimate of polygenic component of variance was 0.192. Thus, the polygenic heritability of the trait, LogTG/HDL is estimated as 0.192/(0.192+0.324)=0.372. About 37% of the total variance can be explained by a polygenic component, suggesting that the trait is highly heritable. The results of testing each marker locus for linkage to an additive major gene influencing the trait, Log TG/HDL were not significant as following table. When we tested for association, there is a polygenic effect and that this may map to the locus where our markers are located, so we must include additive components in the null model. The tests of variance are less powerful than the tests of association, it is possible for an effect to be significant component of the mean but not a significant component of variance. We used orthogonal test of Abecasis et al to evaluate if within-family component of association is significant, and it is robust to stratification. The results showed the most significant result refers to: Marker: ApoA4 (ChiSq: 3.384, p-value: 0.0658). The parametric linkage analysis by LODLINK program did not demonstrated high LOD score among 5 markers.