基于面部分割方法的颅面表型提取及其遗传与进化机制研究

The human craniofacial morphology is the most important distinguishing feature among individuals. The lineage similarity of the face confirms that it is genetically regulated and the heritability of face is about 0.4~0.8. In previous studies, the applicant carried out genetic and evolutionary analysis on several facial observation features and found a series of effective genetic loci. However, craniofacial morphology is a high-dimensional complex feature. The facial phenotyping based on physical description measurement and sparse feature points has limitations because it is typically preselected and has low dimensions. Lack of appropriate phenotypic extraction methods has always been the main reason why craniofacial research cannot be expanded. In response to this problem, this project relies on the sample of 10000-person 3D facial and whole-genome chip data of East Asian population collected in the previous study. On the basis of realizing high-density digital normalization of face model, the applicant will 1) use phenotype or genetic correlation and data-driven method to mine the facial area affected by the same genetic factors, and develop a complete facial segmentation process for dividing relevant regions. 2) extract the craniofacial feature corresponding to the global or local region according to segmentation by dimensionality reduction method, and 3) use genome-wide association scan to locate genetic variation affecting the extracted craniofacial features in East Asian population and conduct the evolutionary study to test whether the loci are selected. This project provides a new idea for the feature extraction of human craniofacial morphology. The results of the study will provide clues for systematic analysis of facial genetic and evolutionary mechanisms. It will also provide evidence for forensic scientists to predict facial morphology from DNA.

颅面外观差异是人类个体间最重要的表型区别。基于家系的研究证实面部相似度的遗传率为0.4~0.8。申请人在前期工作中数字化采集了大量个体的三维面部数据，发现了多个与之显著关联的遗传位点。但人脸差异的构成非常复杂，传统基于稀疏特征点的描述测量表型受人为影响大、数据量小、局限性强，缺乏完善的表型提取方法一直是颅面研究无法拓展的主要原因。针对该问题，本课题拟依托前期采集的东亚10000人三维面部数据和全基因组分型数据，在人脸高维数字化的基础上，①创新利用遗传关联与聚类方法寻找受相同遗传因素影响的颅面区域，制定一套完整的区域分割流程；②通过降维方法提取分割后颅面全局或局部区域对应的数字化表型。③对上述表型在全基因组范围开展遗传进化研究，定位相关遗传变异，检验其进化是否受到选择影响。本项目预期可为人脸的特征提取提供新思路，为系统解析人脸遗传进化机制提供线索，也为法医刑侦利用DNA推测面部形态提供依据。

人类颅面表型是个体间最重要的区别特征，具有高度的遗传性。不同群体的颅面特征有着明显差异，其遗传与进化机制有待挖掘。项目负责人基于前期积累的大样本量面部图像和基因组数据，发挥集成优势，构建了一套自动量化面部分割特征的方法。从遗传与进化角度，解析了人类头面部表型差异热点规律，揭示了部分东亚独特面部形态的遗传作用机制，也定位出了造成欧亚面部差异的遗传因素。具体研究成果包括：1）结合计算机技术与解剖学信息，构建了自动量化面部分割特征的提取流程，加强了生物学可解释性。 2）通过全基因组关联分析定位了244个与头面部特征关联的遗传变异，证实面部特征作为典型的复杂性状由多基因共同影响。3）基于欧亚不同人群面部形态差异，利用多基因形状（Polygenic Shape， PS）方法模拟由遗传因素影响的欧亚特征，进一步定位了13个影响东亚特有面部特征相关的主要遗传因素。4）利用自然选择分析发现欧亚面部受到选择的主要区域为鼻部，并且令鼻部变高变窄的选择主要发生在欧洲而非亚洲。本项目的研究总结并拓展了人脸特征提取方法，为系统解析人脸的遗传进化机制提供了线索。项目的研究成果是以表型组学理念、范式和方法开展生命科学研究的成功尝试，首次阐明了不同人群面部表型差异的潜在遗传基础，大大拓展了东亚面部表型遗传学的知识，对人类自身的认识具有重大意义。

内容获取失败，请点击重试