Chinese genetics of Uyghur face prediction

Four papers by Chinese geneticists have been recently retracted, on the charges that DNA of the oppressed Uyghur minority was used without participants’ consent for purposes of racial profiling.

The idea was to predict Uyghur faces from genetic data for the purposes of mass-surveillance. Two of these Chinese researchers of Uyghur profiling work in Netherlands, one as professor. His Erasmus University explicitly doesn’t mind. Yet another geneticist of racial profiling works in a Beijing partner institute founded by the German Max-Planck Society.

The genetic face prediction tinkering in the West officially aims to serve police forensics, to catch criminals and terrorists. All the state needs to spot your face anywhere in the crowd are some DNA traces of yours they once sampled, how sci-fi is that. Scientists and biotech startups keep announcing progress for many years now, but apparently biological reality keeps getting in the way. Indeed, whether face prediction from genes is actually doable at all is another issue. But who cares if all that is pseudoscientific tinkering: there are enough such forensic science efforts happening in China and also in the West, with a constant stream of peer-reviewed papers claiming success, while the untraceable funding money from governmental authorities flows without restrictions.

In Communist China of Xi Jinping, the purposes are openly sinister, targetting the Uyghur minority, stripping them of their basic human rights, erasing their culture, enslaving them in forced labour camps and now turning them into defenceless subject of scientific experiments aiming to deliver even more oppression. The scheme was exposed by a New York Times investigation already in 2019, when the now retracted peer-reviewed papers first were published.

Academics trust each other blindly and willingly – after all, why wouldn’t China’s Uyghur camp prisoners not willingly sign an informed consent, in a language they don’t speak, threatened with persecution for extremism if they don’t comply? What reason did any academic peer have to suspect any coercion there, right? Were it not for media attention like from the New York Times in December 2019, the two now retracted papers would probably remain untouched.

Four Retractions

These are the papers and their very recent retraction notices:

Yi Li, Wenting Zhao, Dan Li, Xianming Tao, Ziyi Xiong, Jing Liu, Wei Zhang, Anquan Ji, Kun Tang, Fan Liu & Caixia Li EDAR, LYPLAL1, PRDM16, PAX3, DKK1, TNFSF12, CACNA2D3, and SUPT3H gene variants influence facial morphology in a Eurasian population Human Genetics (2019) doi: 10.1007/s00439-019-02023-7

The authors explained who their “Eurasian” sample was exactly:

“This study was approved by the Ethics Committee of the Institute of Forensic Science of China, and all individuals provided written informed consent. The participants were all volunteers. The consent was discussed in their native
language and the signature was in their native language. We sampled a total of 612 unrelated Eurasian individuals living in Tumxuk City in Xinjiang Uyghur Autonomous Region, China. All individuals met the following conditions: (1) their
parents and grandparents were both of Uyghur origin;…“

The Retraction Note from 30 August 2021 went:

“The Editors-in-Chief have retracted this Article. Since publication, concerns were raised about the ethics and consent procedures for this study. We requested supporting documentation from the authors, including the application form submitted to the ethics committee and evidence of ethics approval. The documents supplied by the authors contain insufficient information related to the scope of the study for us to remain confident that the protocols complied with our editorial policies or are in line with international ethical standards (World Medical Association Declaration of Helsinki. Ethical principles for medical research involving human subjects [1]).

Caixa Li stated on behalf of all co-authors that they do not agree to this retraction.”

Another retracted paper was:

Xiaoxi Jing, Yanan Sun, Wenting Zhao, Xingjian Gao, Mi Ma, Fan Liu & Caixia Li Predicting adult height from DNA variants in a European-Asian admixed population International Journal of Legal Medicine (2019) doi: 10.1007/s00414-019-02039-8

The study mentioned:

“We sampled a total of 689 unrelated Uyghur males living in Xinjiang Uyghur Autonomous Region, China. All individuals meet the following conditions: (1) their parents and grandparents were both of Uyghur origin….”

and assured:

“The study was approved by the ethics committee of Institute of Forensic
Science, Ministry of Public Security, China. All participants provided
written informed consent.”

That paper’s retraction notice from 7 September 2021 was worded very similarly to the above, also here all authors disagreed with the retraction.

There were two more retractions for the team of Caixia Li in that journal:

Wei-Qi Hao, Jing Liu, Li Jiang, Jun-Ping Han, Ling Wang, Jiu-Ling Li, Quan Ma, Chao Liu, Hui-Jun Wang & Cai-Xia Li Exploring the ancestry differentiation and inference capacity of the 28-plex AISNPs International Journal of Legal Medicine (2019) doi: 10.1007/s00414-018-1863-z

That retracted paper also used Uyghur DNA collected under coercion:

“Among the samples, 486 individuals from Xinjiang and 117 individuals from Henan and Guangxi Provinces were provided by the National Infrastructure of Chinese Genetic Resources (NICGR:YCZYPT[2017]01-3) […] All subjects provided written informed content and self-declared ancestry information“.

Also here, similar retraction notices were issued and the authors disagreed with the retraction.

"Li Caixia, chief forensic physician of China’s Institute of Forensic Science who obtained DNA samples from [Kenneth Kidd], was at one point a visiting research scientist in genetics, …” #CCP #GeneticProfiling #GeneticsEthics #CloseTheCamps https://t.co/9pvkz3pFoV

— nukinfo (@nukinfo) July 8, 2019

It wasn’t just Uyghurs. The Chinese scientists also racially profiled inhabitants of occupied Tibet, unsurprisingly under coercion and without proper consent:

Li Jiang, Jianxiong Peng, Meisha Huang, Jing Liu, Ling Wang, Quan Ma, Hui Zhao, Xin Yang, Anquan Ji & Caixia Li Differentiation analysis for estimating individual ancestry from the Tibetan Plateau by an archaic altitude adaptation EPAS1 haplotype among East Asian populations International Journal of Legal Medicine (2018) doi: 10.1007/s00414-018–1789-5

The now retracted paper mentions:

“Table S1 summarizes the 4986 individuals from 92 populations used in this study. Blood samples from 1053 individuals from 12 populations were collected by our lab, including 177 individuals from the Tibetan Plateau, 468 individuals from
peripheral regions of Tibet (Qinghai, Gansu, and Sichuan), and 408 individuals from plain regions (Guangxi, Jiangxi, Shandong, and Henan). All of the samples were obtained with informed consent and self-declared ancestry information.“

The paper also includes an acknowledgement which seems to be a form of Chinese Communist party humour:

“Special thanks are due to the hundreds of individuals who volunteered to give blood samples for studies of genetic diversity.“

Dutch concept of Academic Freedom

The Dutch online magazine Follow The Money writes who the authors and their paymasters were:

“Both scientific studies involved researchers working for the Chinese police. One of the authors of the article in the International Journal of Legal Medicine [Mi Ma, -LS] is even affiliated with the police station of the Xinjiang Production and Construction Corps, the paramilitary organization that puts Uyghurs to forced labour in cotton fields in Xinjiang and is involved with the camps where it is estimated that at least a million Uyghurs are imprisoned. For that reason, the US government banned in December 2020 the import of cotton from XPCC.

One of the authors of the two withdrawn studies is Caixia Li, chief forensic scientist of the Ministry of Public Security in China. The research in Human Genetics was funded by the National Laboratory of Forensic Sciences in China. The study in the International Journal of Legal Medicine received money from two forensic labs of the Chinese Ministry of Public Security.”

Another corresponding author is Fan Liu, since 2013 assistant professor at the Department of Genetic Identification (sic!) at the Erasmus University in Rotterdam, Netherlands, which prompted the Follow the Money coverage. Also the co-author Ziyi Xiong proved to be a PhD student at Erasmus University, his 2019-2022 thesis project is about “Epigenetics & lifestyle factors”. Fan Liu main academic job is at the University of Chinese Academy of Sciences in Beijing, his institutional webpage serves this bio:

“he studied and worked at the Medical Center of Erasmus University in the Netherlands for 13 years (2002-2015). He successively studied under Professor Conelia van Duijn, Academician of the Royal Netherlands Academy of Sciences, and Professor Manfred Kayser, an expert in forensic molecular biology. Worked as postdoc, Tenure assistant professor, and doctoral supervisor. In February 2015, he was introduced by the National Youth Award [The Thousand Talents Plan for Young Professionals, awarded 2015-2018, -LS]. In October, he joined the Beijing Institute of Genomics of the Chinese Academy of Sciences full-time and served as a researcher at the Key Laboratory of Precision Genomics Medicine of the Chinese Academy of Sciences.“

Follow the Money asked if the Dutch university has a problem with such research misconduct:

“When asked whether Erasmus MC will take measures against Fan Liu and Ziyi Xiong, the answer is short: ‘No’.“

I contacted Liu’s PhD mentor and head of department at the Erasmus University Medical Center (Erasmus MC), Manfred Kayser. He wrote to me:

“In his work for Erasmus MC there are no irregularities, so yes, we trust his research at Erasmus MC. […]

Dr. Liu’s work in these two retracted articles took place outside his work at Erasmus MC. Dr. Liu did not receive Erasmus MC funding for the work he did outside Erasmus MC that is described in these two retracted articles. Erasmus MC had no knowledge about the work Dr. Liu did outside Erasmus MC that is described in these two retracted articles, until these papers were published in scientific journals. Erasmus MC has no collaboration agreement with the Beijing Institute of Genomics.

Dr. Liu had listed Erasmus MC as his second affiliation after the Beijing Institute of Genomics listed as his first affiliation on these two retracted articles because it is normal in scientific publishing for scientists with multiple affiliations to list all their affiliations on all their publications.“

Interesting. Kayser explicitly states the Erasmus University doesn’t care what their assistant professor did, does or will do in future when not in Holland, and he is free to use his Erasmus affiliation for that. I guess having another job of full-time human rights abuse abroad is their idea of academic freedom?

Kayser also told me that “the Thousand Talents Program of the Chinese government is for recruiting high-level scientists and talents from overseas to China“, underscore his. I then placed a freedom of information inquiry about how Liu’s employment translates into financial support from China to Erasmus University, and will update when they reply.

Erasmus knew of the papers since they appeared in 2019. In the same year, Kayser, Ziyi Xiong and Fan Liu published this paper:

Ziyi Xiong , Gabriela Dankova , Laurence J Howe , Myoung Keun Lee , Pirro G Hysi , Markus A De Jong , Gu Zhu , Kaustubh Adhikari , Dan Li , Yi Li , Bo Pan , Eleanor Feingold , Mary L Marazita , John R Shaffer , Kerrie McAloney , Shu-Hua Xu , Li Jin , Sijia Wang , Femke MS De Vrij , Bas Lendemeijer , Stephen Richmond, Alexei Zhurov, Sarah Lewis, Gemma C Sharp, Lavinia Paternoster, Holly Thompson, Rolando Gonzalez-Jose, Maria Catira Bortolini, Samuel Canizales-Quinteros, Carla Gallo, Giovanni Poletti, Gabriel Bedoya, Francisco Rothhammer, André G Uitterlinden, M Arfan Ikram, Eppo Wolvius, Steven A Kushner, Tamar EC Nijsten, Robert-Jan TS Palstra, Stefan Boehringer, Sarah E Medland, Kun Tang, Andres Ruiz-Linares, Nicholas G Martin, Timothy D Spector, Evie Stergiakouli, Seth M Weinberg, Fan Liu, Manfred Kayser Novel genetic loci affecting facial shape variation in humans eLife (2019) doi: 10.7554/elife.49898

The cohort included a certain “Xinjiang Uyghur Study (UYG, N = 858, Uyghurs from China with ancestry admixture estimated at 50% East Asian and 50% European)“:

“The UYG samples were collected at Xinjiang Medical University in 2013–2014. In total, 858 individuals (including 333 males and 525 females, with an age range of 17–25) were enrolled. The research was conducted with the official approval from the Ethics Committee of the Shanghai Institutes for Biological Sciences, Shanghai, China. All participants had provided written consent.”

Are those perchance the same samples used in the retracted papers? I asked Kayser and Erasmus University and will update. In this regard, Follow the Money references Human Rights Watch who in turn explain why the Chinese authors’ assurances of informed consent are not to be trusted:

“Chinese authorities in Xinjiang are collecting DNA samples, fingerprints, iris scans, and blood types of all residents in the region between the age of 12 and 65, Human Rights Watch said today. […] For all “focus personnel” – those authorities consider threatening to regime stability – and their family members, their biometrics must be taken regardless of age. Authorities are gathering the biodata in different ways. DNA and blood types are being collected through a free annual physical exams program called Physicals for All. It is unclear if the participants of the physicals are informed of the authorities’ intention to collect, store, or use sensitive DNA data.[…]

Local health authorities are responsible for collecting DNA and blood type information “as part of” the Physicals for All program, according to the guidelines. The collected blood type information is directly sent to the police, while the “blood cards for DNA collection will be sent to the county police bureaus for profiling.” All of this information is stored and linked to an individual’s national identification number. […]

One Uyghur who participated in the 2016 Physicals for All program in Kashgar in western Xinjiang told Human Rights Watch that his neighborhood committee “had demanded that they [people in his neighborhood] must participate in the physicals.” He did not think he had a choice in the matter, as “not participating would surely be seen as a sign of ‘thought problem,’” a shorthand for “political disloyalty,” a dangerous label in the repressive region. He said the health authorities had not told him afterward the results of his physical.“

Back in 2019, New York Times reported:

“The police prevented reporters from The New York Times from interviewing Tumxuk residents, making verifying consent impossible. Many residents had vanished in any case. On the road to one of the internment camps, an entire neighborhood had been bulldozed into rubble.”

Imagine the gripping stories Fan Liu can be telling his Dutch mentor when back in Rotterdam! Not your everyday field trip!

How to recognise Uyghurs by the ears

One collaborator, never featuring as co-author but often in acknowledgments for providing technical tools, was Kenneth K. Kidd, emeritus professor at the Yale School of Medicine. His involvement was exposed by the New York Times investigation already in 2019, back then Kidd claimed to have been a naive victim of Caixia Li’s scheming, tricked into collaborating in 2010. Yet this article by Katherine Hu was sceptical of Kidd’s evasive stance, already in September 2019:

“ALFRED [Allele Frequency Database] has been Kidd’s pet project since 1999. Currently, it contains data on DNA samples from 2,942 Uighurs, most of which were collected in China over the last two decades. […] Kidd continues to add new populations to ALFRED; his most recent paper, released after the controversy, examines twenty-five new ethnic groups for forensic markers, adding their population samples to ALFRED. Among those are the Chengdu Tibetans, Liangshan Tibetans, Qinghai Tibetans, and the Yi, all of which are ethnic minorities in China.”

There are of course more racial profiling studies by Caixia Li, Fan Liu and others. There are also papers in Chinese, and there is no hiding anymore behind “Eurasian” circumscription. Did Erasmus MC know about these also? The works sure carry Liu’s Dutch affiliation. Like this one:

Li Yi, Zhao Wenting, Li Dan, Tao Xianming, Xiong Ziyi Liu Jing, Zhang Wei, Liu Haibo, Tang Kun, Liu Fan, Li Caixia The effect of EDARV370A on facial and ear morphology of Xinjiang Uyghur population Hereditas (Beijing) (2018) doi: 10.16288/j.yczz.18-268 (PubMed abstract)

Here the abstract:

“The ectodysplasin A receptor gene (EDAR) plays an important role in the development of ectoderm. The derived G allele of its key missense variant EDARV370A is prevalent in East Asians and Americans, but rare in Africans and Europeans. This leads to distinct ectodermal-derived phenotypes between different continental groups, such as the straighter and thicker hair, more eccrine sweat glands, feminine smaller breasts, shovel incisors characteristic of East Asians. At present, we know little about the association between EDARV370A and facial and ear morphology characteristics. To better understand the effect of EDARV370A on craniofacial phenotypes, we systematically examined the association between EDARV370A and 136 facial quantitative phenotypes, one chin ordinal phenotype and six ear ordinal phenotypes in 715 Uyghurs. The quantitative phenotypes were derived by applying our automated landmark annotation method to facial 3D photos and the ordinal phenotypes were manually graded from facial 2D photos. The analysis identified significant association (P<0.05 after multiple testing correction) between EDARV370A and eight facial phenotypes, one chin phenotype and three ear morphology phenotypes. Our study thus elucidated the pleotropic effect of EDARV370A on craniofacial phenotypes in a European-Asian admixed Uyghur population.”

Max Planck Society Partners

One of the key contributors here, next to Fan Liu and Caixia Li, is Tang Kun. He was also co-author on one of the four retracted papers and also on the aforementioned eLife paper with Kayser, Liu and Xiong at Erasmus MC. Professor Tang is affiliated with the Partner Institute for Computational Biology (PICB) of Chinese Academy of Sciences-Max Planck Society in Shanghai. This is also probably why the ethics approval for the elife data came from Shanghai.

Tang’s German collaborator (according to this 2014 annual report) is professor Mark Stoneking of the Max Planck Institute for Evolutionary Anthropology in Leipzig. Stoneking however informed me:

“I am not a co-author, collaborator, or in any way involved with that specific study and so I have no knowledge concerning the study population or informed consent procedures used in that study.“

The Max Planck Society also announced a statement, I will update when it arrives. The New York Times reported in this regard in 2019:

“Dr. Tang said he was unaware of the origins of the DNA samples examined in the two papers, the 2018 paper in Hereditas (Beijing) and the Human Genetics paper published in April. The publishers of the papers said they were unaware, too. Hereditas (Beijing) did not respond to a request for comment.”

Here another such paper by Kun Tang with Caixia Li:

LIU Jing, QIAO Lu, ZHAO Wenting, JIANG Li, JI Anquan, WANG Guiqiang, YE Jian, TANG Kun, LI Caixia Experimentation of Human Facial Predication by Relevant DNA SNPs DNA Forensic Science and Technology (2017) doi: 10.16467/j.1008-3650.2017.04.002 

“Through selection of 350 facial morphology-related SNPs analysed from sequencing 24 Chinese males (18 Uyghur and 6 Han), the relevant SNP phenotypes were obtained so that a model of facial morphologic prediction was built up based on such one of the previously-developed. By evaluation of the similarity between the genetic predicting faces and the real ones to show the model’s accuracy, some problems of the current technique were discussed together with the further researches that will be inevitably carried out in the future.“

Back in 2019, the Max Planck Society was asked by The New York Times to comment, and were told:

“The German organization also provided $22,000 a year in funding to Dr. Tang because he conducted research at an institute affiliated with it, said Christina Beck, a spokeswoman for the Max Planck Society. Dr. Tang said the grant had run out before he began working with the police, according to Dr. Beck. […] Dr. Beck, the Max Planck spokeswoman, said Dr. Tang had told the organization that he began working with the police in 2017, after it had stopped funding his research a year earlier.“

In 2019, Tang also informed NYT:

“In an interview, Dr. Tang said he did not know why he was named as an author of the April paper, though he said it might have been because his graduate students worked on it. He said he had ended his affiliation with the Chinese police in 2017 because he felt their biological samples and research were subpar.

“To be frank, you overestimate how genius the Chinese police is,” said Dr. Tang, who had recently shut down a business focused on DNA testing and ancestry.”

Basically, he never really did any Uyghur sampling. Those papers? Nothing to do with Professor Tang.

Update 16.09.2021 The Max Planck Society spokesperson Christina Beck informed me:

“The contractual relationship with the CAS-MPG Partner Institute for Computational Biology PICB has expired since 31.12.2020. Since 1.1.2021, this name may also no longer be used. The CAS has informed us that the institute in the previous form actually no longer exists, the groups have been transferred to the CAS Shanghai Institute of Nutrition and Health.“

Secret grants

And now Caixia Li and her team fine-tuned their face prediction of Uyghur victims by adjusting for hair loss:

Pan Siyu, Zhao Wenting, Feng Rui, Li Qiong, Jing Xiaoxi, Gao Xingjian, Chen Yan, Li Yi, Tao Xianming, Liu Tianzi, Li Caixia Predicting male pattern hair loss in a mixed population of Asia and Europe based on DNA variation Progress in Biochemistry and Biophysics (2020) doi: 10.16476/j.pibb.2020.0095 

“This study is based on 684 Eurasian samples from a mixed population of Asia and Europe in China, and 624 MPB-related SNPs found in the UK Biobank GWAS analysis of 205 327 European males were analyzed for population heterogeneity, based on polygenic risk scores…“

You know who the “Eurasian” samples were.

The Chinese grants acknowledged in the papers are not available in any official national data bases. Like this, from the retracted Li et al 2019:

 “This project was supported by the National Key R&D Program of China (2017YFC0803501), National Natural Science Foundation of China (91651507), Fund from institute of forensic science (2018JB046), and open projects of the National Engineering Laboratory for Forensic Science (2017NELKFKT05). Biological samples were provided by the National Science and Technological Resources Platform (YCZYPT[2017]01-3 and 2017JB025). Author FL [Fan Liu] is supported by “The Thousand Talents Plan for Young Professionals”. Author CXL [Caixia Li] is supported by “The Beijing Leading Talent Program (Z18110006318006)”.

While the grants are hidden because of politics and secrecy, the papers referencing these grants can be found. Here for example, a list of dozens of papers with the grant 2017YFC0803501, all on some kind of racial profiling, and almost all are related to the Ministry of Public Security or local law enforcement.

But then again, who cares. The future of science is in China, and western academics and universities are too eager to follow.

---