基于基因序列相似性的分子序列聚类可以提供物种数量和多样性的信息。该方法被广泛用于评估很多类群物种水平的多样性的研究中。目前，尽管多基因数据迅速增长，现有物种界定的文献资料中仍以单基因分析为主导，很少讨论将多位点的数据集整合到物种聚类工作中。 Douglas Chesters博士今年完成了在我研究组（zhucd@ioz.ac.cn）为期两年的博士后研究工作，并顺利成为助理研究员。他与多个研究机构研究人员合作，研究出了一种新的基于多基因序列数据进行物种聚类的启发式搜索方法。该方法首先完成单个基因的物种聚类，然后用一种新颖的方法进行整合，并在搜索过程中寻找并优化最佳参数。该研究工作测试了许多已有数据集，并对一个蜜蜂样本的三基因的新测数据集进行分析。这些蜜蜂样本采集自北京昌平南口一个苹果种植园及周边山沟。以图1为例，本研究基于3个基因片段信息，聚类了250个蜜蜂样本，界定了70个物种，并在对部分样品实现自动分类标记，实现物种鉴定。作者论证了该方法对中型到大型多基因数据集估算物种多样性研究中的实用性。同时该工作也发现在物种聚类过程中，某些保守基因如28S基因与其它基因或形态学数据分析结果存在较大的分歧。 该研究符合物种多样性调查向多基因研究方法的转变趋势，并提出了新颖的方法。该文章已经被《Methods in Ecology and Evolution》接收，并在线发表（论文链接：http://onlinelibrary.wiley.com/doi/10.1111/2041-210X.12104/abstract）。  图1：一种新的多基因数据集的物种聚类方法在北京蜜蜂序列的数据集上的应用。每一个扇形图是一个单独的物种单元，其大小与序列组成的物种单元的数量成比例。颜色代表基因片段，蓝色=CytB，白色=COI，红色=28S。结果显示共有70个物种单元。
Heuristic optimization for global species clustering of DNA sequence data from multiple lociClustering molecular sequences based on the level of similarity can give information on the number and diversity of species, and is a very commonly used approach for assessing species level diversity in many taxonomic groups. The number of available multi-gene datasets is growing rapidly, although the literature is currently dominated with single gene analyses, with remarkably little discussion of integrated species clustering using multi-locus dataset. Dr Douglas Chesters finished his 2-year postdoc in Prof. Chao-Dong ZHU’s lab (zhucd@ioz.ac.cn) and got a position as the assistant professor on molecular systematics in the same lab at the Key Laboratory of Systematics and Evolution (CAS), Institute of Zoology, Chinese Academy of Sciences. He collaborated with colleagues in the same and other institutions, reporting a new heuristic approach for clustering multi-gene sequence data at the species level. The approach uses standard species clusters at each gene under study, but these are then combined in a novel way, which looks for parameters producing the best set of combined species entities during a search procedure. The authors use a number of test datasets, including a newly sequenced 3-gene dataset of bee specimens collected from an apple orchard in Beijing. Figure 1 shows an example, with the results for the clustering of three genes and 250 bee specimens, with the method estimating 70 species units, and giving automated taxonomic labels in some cases. The authors demonstrate the usefulness of the approach for estimating species diversity of medium to large multi-gene datasets, although finding high disagreements between the species hypotheses formed particularly at the 28S gene.The study should become quite relevant as surveys of species diversity switch to multi-gene methods. The paper has been accepted and on-line by Methods in Ecology and Evolution (http://onlinelibrary.wiley.com/doi/10.1111/2041-210X.12104/abstract). Figure 1: A novel approach to speciesclustering of multi-gene datasets applied to sequence data from Beijing bees. Each pie chart gives a single species unit, with the size proportional to the number of sequences comprising the species unit. They are colored according to genes represented, where blue segments = CytB, white = COI, red= 28S. The results suggest 70 species units in total. Douglas Chesters博士（dchesters@ioz.ac.cn）提供了该文章的英文简介，刘秀薇博士提供了中文简介。 |
