Computer assisted fish karyotyping

ABSTRACT Fish karyotyping plays an important role in studies of cytotaxonomy and chromosomic evolution of species. However, few semi or completely automated fish karyotyping systems are available. This work proposes and evaluates an image-based tool to assist fish karyotyping. The analyzed species were Hopliasmalabaricus (Bloch, 1794), Hypostomusancistroides (Ihering, 1911), and Parauchenipterusgaleatus (Linnaeus, 1766). In Portuguese, these species are commonly referred to as traíra, cascudo, and bagresapo, respectively. A total of 100 metaphases were analyzed through two methods: (1) semi-automatic karyotype generation and (2) automatic karyotype generation. The results were analyzed using Pearson correlation, difference graphs and counting tables. The reference used for the evaluation of the system was the average of the counts made by four experts. In method 1, four users performed counts with interobserver correlation of r≥ 0.997. The total number of chromosomes identified by method 1 was 4348 and method 2 was 4135, excluding false positives, resulting in an automatic identification of approximately 95,1% of the chromosomes, resulting in a correlation between the methods of r= 0.93. The results indicate that the tool can be introduced for fish karyotyping procedures contributing for accelerating the process with acceptable accuracy.

摘要：鱼类核型分析（karyotyping）在物种细胞分类学（cytotaxonomy）与染色体进化（chromosomic evolution）研究中具有重要作用。然而，当前可供使用的半自动化或全自动化鱼类核型分析系统仍较为匮乏。本研究提出并评估了一款用于辅助鱼类核型分析的图像工具。本次分析涉及的物种为马拉巴虎脂鲤（Hoplias malabaricus，Bloch, 1794）、钩棘甲鲶（Hypostomus ancistroides，Ihering, 1911）以及盔副管鲶（Parauchenipterus galeatus，Linnaeus, 1766）；在葡萄牙语中，这三个物种的常用俗称分别为traíra、cascudo与bagresapo。本研究通过两种方法共分析了100个分裂中期细胞（metaphase）：（1）半自动核型构建法，（2）全自动核型构建法。研究采用皮尔逊相关分析（Pearson correlation）、差异统计图与计数表格对结果进行分析，并以4位专家的计数结果平均值作为系统评估的参照基准。在方法1中，4名操作人员完成了染色体计数，观察者间相关系数（interobserver correlation）r≥0.997。剔除假阳性（false positives）结果后，方法1共识别出4348条染色体，方法2共识别出4135条染色体，自动识别准确率约为95.1%；两种方法的相关系数r=0.93。研究结果表明，该工具可应用于鱼类核型分析流程，在保证可接受精度的前提下加快分析进程。