Statistical pattern analysis of D1S80 alleles in Northwestern Russians and worldwide database using COLLAPSE software
详细信息 查看全文
- 作者：Smolyanitsky ; A.G. ; Khromov-Borisov ; N.N. ; Popov ; V.L. ; Zaslavsky ; G.I. ; Rogozin ; I.B. ; Henriques ; J.A.P. ; et. al.
- 关键词：Forensic population genetics ; Databases ; Statistical analysis
- 刊名：International Congress Series
- 出版年：2003
- 出版时间：January, 2003
- 年：2003
- 卷：1239
- 期：Complete
- 页码：665-671
- 全文大小：154 K
文摘
Polymorphism at the locus D1S80 was studied in 283 unrelated habitants of the Northwestern Federal Region of Russia that includes the cities Arkhangel'sk, Kaliningrad, Murmansk, Novgorod, Petrozavodsk, Pskov, Saint Petersburg and surrounding districts. The sample distribution obtained was compared with the published data on 128 worldwide samples from the DNA-PCR Databank. For this purpose a new statistical technique—Similarity Pattern Analysis (SPAN) and corresponding COLLAPSE software were used and their validity and applicability were demonstrated. The main features of the method are the following: First, to measure the similarity (homogeneity) between any pair of populations in the study, the relevant sufficient statistics, Kastenbaum–Hirotsu squared distance (KHi2), is used. Second, the method is based on the collapsing principle, which permits similar subsets of the sample distributions to be combined (to collapse) into distinct (locally homogeneous) groups (homoclusters). Such a procedure permits the most likely (optimal) version of the similarity pattern to be revealed. Thirdly, the discrimination among homo- and hetero-clusters is based on the so-called χ2Reduction Principle, according to which the corresponding statistics, χ2reduction (CSR), as a measure of intra-cluster homogeneity should be kept non-significant after each step of collapsing. Evaluated similarities and/or dissimilarities appeared to be rather reasonable and interpretable.