發展連鎖STR標記以提升親緣關係鑑定準確率

Abstract

短片段重複序列(Short Tandem Repeat, STR)是由2到7個鹼基的序列不斷重複所構成的短片段DNA，其被廣泛應用在法醫學上之人身鑑別和親緣關係鑑定。目前一般常用的STR基因(loci)多選自美國DNA整合索引系統(Combined DNA Index System, CODIS)所公布的13個基因；市面上大部份的商業鑑定組也多涵蓋這些基因。 然而，在親緣關係鑑定的實務上常會發生只有少數幾個基因不相符的情形，在此法完全排除或確認二者之親緣關係的情況下，要考慮此一結果是否是突變所造成；此時一般常用的核心基因並不一定能提供足夠的鑑別力，也因此常需要額外的基因來增加鑑別能力。目前許多新發展出的商用鑑定試劑針對這些問題提供額外的鑑定服務，這些鑑定組合多針對與核心基因沒有連鎖的基因。然而其過程費時又費工且計算繁雜，再者，若是新增加之基因也發生突變時此一問題會更為複雜。本研究之主要目的為在常用核心基因的前後尋找與其連鎖(linked)之新STR基因，如此在少數幾個核心基因不相符時，只要確認此一連鎖之基因型是否相符，即可初步研判此一不相符的原因是否為突變所造成。 本研究在十三個常用之核心基因(TPOX、D2S1338、D3S1358、FGA、D5S818、CSF1PO、D7S820、TH01、VWA、D16S539、D18S51、D19S433及D21S11)的前後各20萬個鹼基序列中搜尋新的連鎖STR標記，分別找到了相對應的13個以四鹼基重複序列的新STR標記，並以在原本的基因名稱後方加入小寫英文字母a暫定為其名稱。分析這些可能的新STR標記在台灣230個無親緣關係之漢人檢體中的型別與其分布頻率，發現除了TPOXa僅三個對偶基因(allele)型別，以及D3S1358a與D16S539a僅四個型別外，其餘的新STR基因都至少有七個型別。再進一步分析其實用性，發現除了D16S539a外這些新STR標記的多型性資訊含量(polymorphism information content, PIC)值皆大於0.44，達到法醫學上評估標記之標準；而異質性(heterozygosity)則僅D2S1338a未達法醫學上之標準0.5。此外，D5S818a、CSF1POa及D7S820a的各項數值皆超越其相對應之核心基因。 連鎖分析(linkage analysis)則顯示這些新STR基因和其相對應之核心STR基因重組率低，確實有連鎖的情形。再進一步分析這些STR標記在族群中的分布是否有偏離哈溫平衡(Hardy-Weinberg equilibrium)及連鎖不平衡(linkage disequilibrium)的情形，可發現D5S818a和D21S11a均有顯著偏離哈溫平衡的情形(p < 0.05)；而連鎖不平衡亦可以在六個STR組合(D2S1338、D3S1358、FGA、D7S820、VWA、D21S11)觀察到(p<0.05)。然而若是使用家族檢體進行分析，則僅TPOX/TPOXa及TH01/TH01a兩對STR組合沒有連鎖不平衡的情形。我們推測造成上述結果最有可能之原因為分析之檢體數量太少或是此兩新發現之STR標記其等位基因(allele)數較少，若是能擴大母群體的樣本數、並加入不同人種的檢體進行分析，應可進一步確認其原因。

STRs (short tandem repeats) have been the primary tool in forensic DNA testing for decades, they are widely used in many jurisdiction systems for kinship analysis. Different countries and jurisdictions have different set of STR core loci, but most of them followed the loci used in US database CODIS, and most commercial kits would include these loci. Sometimes the STR genotyping results are inconclusive, it can either be due to mutations or that the test subjects are distant relatives, such cases often require the addition of more loci. However, more loci would mean more work to be done and much more complicated calculations, and each additional loci would bear its own risk of mutation. Here we offer a simpler approach: by finding new STR loci closely linked to the core loci commonly in use, in cases only a small number of loci are incompatible, we can identify if the reason behind this is really mutation by comparing the genotype of linked STR markers. A total of 13 new STR markers were find within close range from core loci (TPOX, D2S1338, D3S1358, FGA, D5S818, CSF1PO, D7S820, TH01, VWA, D16S539, D18S51, D19S433 and D21S11), they are tentatively named by adding a to the name of original core loci. Most of these new STRs have more than 7 alleles, except for TPOXa (which has 3), D3S1358a and D16S539a (both have 4). As for the usefulness of these new markers, with the exception of D16S539a, the PIC (polymorphism information content) of all markers exceeded 0.44; and the heterozygosity of most loci exceeded 0.5 except for D2S1338a. D5S818a, CSF1POa and D7S820a showed better PIC, heterozygosity, matching probability, power of discrimination and power of exclusion compare to their respective core loci. Linkage analysis showed all new STR loci are strongly linked to the core loci. Out of all newly identified STR, only D5S818a and D21S11a showed deviation from Hardy-Weinberg equilibrium (p<0.05), the possible reason behind this is inadequate sample size; 5 STR pairs showed linkage disequilibrium with unrelated population samples, however, using family samples, only TPOX/TPOXa and TH01/TH01a did not show linkage disequilibrium (p>0.05), we expect better understanding of the true reasons behind this once more samples are involved.