Seeing Faces and History Through Human Genome Sequences
Total Page:16
File Type:pdf, Size:1020Kb
SPECTRUM Seeing Faces and History through Human Genome Sequences CAS/MPG Partner Group on the Human Functional Genetic Variations Shanghai-Leipzig, 2011.2.1–2016.1.31 Prof. Dr. TANG Kun (middle) with his cooperator, Prof. Dr. Mark Stoneking (right) and colleague, Prof. Dr. LI Haipeng (left). The scientific goals of the Partner Group are to utilize large scale genomic polymorphism data to make fine inferences about human demographic history and various forms of natural selection in the human genome; and to understand the genetic basis and evolutionary mechanisms underlying the common variations in human facial morphology, both within and between populations. Reported by Group Leader TANG Kun Photos/Images: By courtesy of Prof. TANG Kun. Photos/Images: By courtesy of Prof. TANG Networking for Science 161 BCAS Vol.28 No.2 2014 Figure 1: Detecting genetic association of common facial variation with high density 3D image registration: SNP rs642961, a genetic risk factor of NSCL/P, strongly predisposes the mouth shape in healthy Han Chinese females. Top: pixel-wise distance as heat-plot be- tween the average facial shapes of wild types (CC) and mutants (TT). Bottom: mouth image projections on a discriminant hyper-line, where TT differ most from the other two geno- types. Middle: the modulation of rs642961 TT on mouth shape was simulated as 3D model (front) via linear transformation along the hyper-line, compared to the model simulating the opposite effects (behind). See Peng et al. 2013 2. Studies of natural selection in hu- questions in one shot, including con- ONGOING RE- man genome trolling the effects of non-equilibrium For natural selection, we are inter- demographics, detecting the loci of var- search ACTIVI- ested in two fundamental questions: ious natural selection scenarios (posi- ties First, how did the genomic signals of tive, negative and balancing selection), natural selection fit into the big picture as well as the consequential parameter of recent human history; second, how estimation, such as the selection times do the footprints of ancient selection af- and selection coefficients for the recent Section 1: Demographics and fect people’s life today. Following these positive selection events. The main idea Natural Selection in Recent Human directions, several projects were carried is to first re-construct the coalescent Evolution out, including characterization of the trees across the whole genome, based global patterns of genetic diversity and on the pairwise coalescent estimation 1. Demographic inference based on signals of natural selection for human via PSMC approach (Li, H., & Durbin, analysis of IBD/IBS ADME genes (In collaboration with the R. (2011). Nature, 475(7357), 493–496). Demographics inference has always group led by Prof. Dr. Stoneking), detec- Using the tree data, several likelihood been an essential research question in tion of signals of recent co-evolution in tests were developed to collectively as- population genetics and is crucial for re- human populations as well as functional sign all genome fragments to modes of liable detection of signals of natural se- characterizing the genetic variants that neutral, negative, balancing or posi- lection. We carried out a series of studies carry strong evidence of recent positive tive selection, and simultaneously es- to find sufficient statistics that possesses selection. timate the key selection parameters. simple mathematical relationships with Interestingly, for the first time we are the demographics, so that demograph- 3. The fine atlas of natural selection able to estimate the selection times for ics can be inferred with high resolution in human genome all the positive selection events in the and power. One set of novel statistics To identify and accurately charac- genome. The temporal resolution of we developed is based on the haplotype terize the genomic loci of natural se- these selection events will shed light extension, which can be used to infer lection is a core problem in population on the question as to what kinds of ad- demographic parameters of relatively genetics. Collaborating with Dr. Ros- aptation events corresponded to agri- complex models (Theunert et al, 2012) tislav Matveev (MPI-EVA), we try to culture expansion (~ 10,000 years ago), (In collaboration with the group led by establish an integrated computational migration and ice-age (20,000 ~ 30,000 Prof. Dr. Stoneking). framework that addresses all the major years ago). 162 Networking for Science SPECTRUM Figure 2: A. the point-wise Qst between Han Chinese and Europeans. The highest values on nasal tip, central eyebrows and cheeks (marked in stars) are 0.48, 0.46 and 0.58 respectively. B. the nose shape divergence on PC modes. PC2 almost completely separates Europeans (green) from Han Chinese (black). The pair-wise Qst on PC2 is 0.79. Section 2: Genetic Basis and Evo- 2. Detecting genetic association of skulls found low levels of divergence, lutionary Mechanisms Underlying common human facial morphologi- therefore supported neutral drift as the the Common Variations of Human cal variation major evolutionary force in the recent Facial Morphology In this study, we applied the dense evolution of human facial morphology. 3D face registration mentioned above In this study we analyzed high-resolu- The human face plays an essential to detect genetic associations. Different tion 3D images of soft-tissue facial sur- role in everyday life. It hosts the most phenotype data schemes were evaluated face in four Eurasian populations: Han important sensory organs and acts as in detecting the potential association of Chinese, Tibetans, Uyghur and Europe- the central interface for expression, mu- facial shape variations. The dense regis- ans. The dense 3D face registration meth- tual identification, and communication. tration scheme showed slightly higher od allowed the facial shape diversity to We would like to develop new technolo- statistical power and stronger advantage be examined at unprecedented resolu- gies and do genetic studies to promote in the fine inference of shape changes tion. We found that noses, cheekbones the understanding of the genetic bases and 3D face modeling. One genetic vari- and eyebrows exhibit strong signals of and evolution of human faces. ant, the rs642961-T in the gene IRF6, a divergence (Qst estimates: 0.3~0.8) be- known risk factor of non-syndromic cleft tween Europeans and Han Chinese (see 1. Automatic landmark annotation lips/palates, was strongly associated Figure 2). The highest divergence rate and dense correspondence registra- with more protrusive lips in the healthy approaches that of skin pigmentation tion for 3D human facial images female Han Chinese (see the figures be- (~0.8). These results therefore suggest We implemented a fully automatic low). Our method opens the possibility that some facial features likely under- and highly efficient 3D face image regis- to systematically scan the subtle impacts went strong selection, either local ad- tration platform that features automatic of genetic variants on facial morphology aptation or sexual selection or both. (In 3D landmark registration and high (Figure 1) (Peng et al., 2013). collaboration with Prof. Dr. Stoneking’s density registration. Furthermore, this group). method is superior in accuracy and effi- ciency. With this new tool, we are able to 3. Variation and signatures of selec- 4. Genome-wide association stud- carry out fully quantitative analyses on tion on the human face ies of common facial morphological thousands of 3D facial images of differ- Divergence of human facial shape changes within and between popu- ent genders, ethnicities and ages. across the globe has been a long-debat- lations ed question. Previous studies based on We are using pooling-based GWAS Networking for Science 163 BCAS Vol.28 No.2 2014 technology to identify genetic loci as- liably re-constructed based on sequenc- INFO sociated with different facial shape ing data. The coalescent trees essentially variations. The main procedures include hold the maximum information one can Group Leader: Prof. Dr. TANG Kun identification of highly variable/diverse achieve from the genetic data about the Founding Date: Feb 2011 features, classification of feature groups, past. The well-defined statistical mod- Max Planck Institute for Evolutionary subsequent DNA pooling based GWAS els of coalescent suggest that tests can Anthropology and the final re-validation of the candi- be designed with moderate ease to ex- Prof. Dr. Svante Pääbo date association signals using single SNP amine all kinds of hypotheses, such as Prof. Dr. Mark Stoneking typing. For example, we observed strong admixture, ancient admixture, migra- CAS-MPG Partner Institute of Computatio- divergence in nose shapes between Han tion and disease evolution, etc. We will nal Biology Chinese and Europeans as described try to explore these questions based on Prof. Dr. JIN Li in study 3. To identify the genetic vari- our new framework of coalescent re- Scientific Advisory Board Members ants underlying such differences, we construction. Prof. Dr. Mark Stoneking (MPI-EVA) choose Uyghur, the admixed population Prof. Dr. JIN Li (Fudan University) between Europeans and East Asians. Direction 2: The Genetic Bases and Prof. Dr. Manfred Kayser (Erasmus MC) Uyghur noses were classified into two Evolution of Human Facial Morpho- Internet Addresses groups of European-like and Han-Chi- logical Diversity http://www.picb.ac.cn nese-like using the PC mode that distin- Phone: +86-21-54920277 guishes the European and Han noses the Our current studies indicate that [email protected] best. Pooling based GWAS was carried methods of high-dimensional data de- out on the two groups and several can- composition are critical to the success- didate loci were identified and partially ful detection of complex and subtle STAFF re-validated in different groups. facial shape differences. Therefore we would put strong efforts into the devel- GUO Jing (Research Assistant) opment of new methods suitable for the LIU Chen (Technician) high dense face data analyses.