在臺灣土雞L2品系比較使用單核苷酸多型性標識和單倍型標識的效力

Abstract

單倍型是染色體區域上的一組傾向於一起遺傳的單核苷酸多型性，比起單個單核苷酸多型性，單倍型可能可以整合一個基因上不同功能的多型性。此外，單倍型可同時與幾種具有功能的多型性有關聯，以減少統計檢測中所需的標識數量，因此可以增加這些檢測的效力。本研究的目的在於評估一個基因上單核苷酸多型性之間的連鎖不平衡之後重組單倍型，並比較使用單核苷酸多型性和單倍型作為遺傳標識的有效性。 自1980年代至今，臺灣土雞L2產蛋品系已選育產蛋數超過三十代。本研究使用臺灣土雞L2品系437隻。其中168隻雞具有40週齡總產蛋數的育種價估計值，使用於關聯性分析。在先前的研究中發現在L2產蛋品系和B產肉品系之間，有29個基因的非同義單核苷酸多型性具有顯著不同的基因型頻率。我們挑選其中11個與脂質代謝有關的基因，並在這些基因設計了96個單核苷酸多型性。在每個基因中，每距離1000至3000個鹼基對選一個單核苷酸多型性位點以覆蓋一整個基因。使用96-plex GoldenGate® assay進行基因型鑑定，剩下435隻雞的65個單核苷酸多型性的結果做後續的分析。在重建單倍型之前，我們估計一個基因上單核苷酸多型性之間的連鎖不平衡以確保具有足夠且有效的連鎖。共有11個基因上的49個單核苷酸多型性具有足夠連鎖不平衡（≥0.77），使用兩種不同的軟體（PHASE和SHAPEIT）將其重組成單倍型。最後，使用變異數分析和逐步迴歸分別分析單核苷酸多型性和單倍型與育種價估計值之間的關聯性。 在一個基因，距離適當的單核苷酸多型性之間若有強的連鎖不平衡則可重組單倍型。我們成功地在11個與脂質代謝相關基因重組57個單倍型。四十週齡總產蛋數的育種價估計值和單核苷酸多型性之間，在基因的層次上並沒有顯著的關聯性。相較之下，在5個基因上有10個單倍型與育種價估計值具有顯著相關。這5個基因的10個單倍型共可解釋33％的育種價估計值變異。由於一個基因上的單倍型是種交替基因效應，具有可相加性，因此我們可透過尋找其他也有顯著效應的單倍型逐步累加其效應：那些對EBV有正向影響的單倍型可用於選拔，負向影響的單倍型可用於剔除。總結而言，在四十週齡總產蛋數之育種價估計值的關聯性分析，單倍型比單核苷酸多型性更有效果。在數量性狀的標識輔助選拔中，單倍型可能比單核苷酸多型性更適合用於遺傳標識。在未來，也可將單倍型推展到全基因體關聯分析進行研究。

Haplotype is a cluster of SNPs on a region of chromosome tending to be inherited together. Haplotypic information might integrate effects of different SNPs on physiological function rather than using single SNP in a region. Haplotype could link with several functional polymorphisms simultaneously to reduce the number of markers needed in the statistical tests, so that may increase the power of those tests. The objective of this study is to estimate linkage disequilibrium (LD) in reconstructing haplotypes and to compare the effectiveness of utilizing SNP and haplotype as genetic marker. The L2 line of Taiwan country chicken has been selected for egg production since 1980’s. There were 437 chicken of two generations in L2 line genotyped for haplotype reconstruction. Among them, 168 chicken with estimated breeding value of egg number laid up to 40 weeks (EBV of EN40) were used in association analysis. In the previous study, non-synonymous SNPs from 29 genes were found with significantly different genotypic frequencies between egg production L2 line and meat production B line. Among them, 11 genes with 96 SNP sites associated with lipid metabolism were studied. In each gene, a SNP site was selected per 1,000 to 3,000 base pairs to cover the whole gene. After genotyping with 96-plex GoldenGate® assay, there were 435 chicken with call rate > 0.9 and 65 SNPs with minor allele frequency > 0.01. Before reconstructing haplotype, we estimate LD between SNPs in a gene to verify their linkage and informativity. Forty night SNPs of 11 genes with sufficient LD (≥ 0.77) were retained to reconstruct haplotypes using two different software packages – PHASE and SHAPEIT. Finally, the association between markers and those EBVs was analyzed by using ANOVA for SNP and stepwise regression for haplotype. SNPs within an adequate distance with strong LD could be reconstructed into haplotypes. Totally, we successfully reconstructed 57 haplotypes in 11 genes. There is not significant result in the association between SNP and EBV of EN40. In contrast, there are 10 haplotypes in 5 genes significantly associated with the EBVs. These haplotypes in the 5 genes might explain 33% of EBV variation. Because the allelic effect of haplotype is additive, we might continually find other significant haplotypes to accumulate their effects in a particular population: those haplotypes with positive effect on EBV could be applied to selection, and the negative ones could be used to cull. To sum up, in association with EBV of EN40, haplotype is more powerful than SNP, and might be potentially better genetic marker than SNP in marker-assisted selection for quantitative trait. In the future, it might be feasible to further exploit haplotype in genome-wide association study.