Not only the external similarity
Geneticists have found similarities in the DNA of unrelated twins
Irina Grishchenko, N+1
Spanish scientists have found out that the strong similarity of people's faces is associated with DNA features: half of the studied pairs of twins had absolutely identical sequences in the genome. The researchers also collected data on the epigenome and microbiome of the mouth — but there was no noticeable similarity. The work was published in the journal Cell Reports (Joshi et al., Look-alike humans identified by facial recognition algorithms show genetic similarities).
People recognize and remember each other, relying on the uniqueness of faces. But sometimes we meet people who are very similar to our acquaintances. Entertainment is even built on such a coincidence, for example, a show of doppelgangers, and it seems an amazing coincidence. At the same time, such people are not related by kinship ties that could explain their similarity. In addition to genes, other mechanisms can affect people's appearance, for example, epigenetic ones — the appearance or disappearance of chemical labels that regulate gene expression. As well as the composition of bacterial communities on different surfaces of the body. But it is not completely clear to scientists what contribution genetics makes to the formation of external signs, and what contribution the environment makes.
Scientists from the Josep Carreras Leukemia Research Institute, led by Manel Esteller, decided to test this on twins. The researchers invited 32 pairs of models from the works of Canadian photographer Francois Brunel, who has been collecting photos of doubles around the world since 1999. Then the scientists objectively determined the measure of similarity of the pairs using a set of face parameters and three different facial recognition algorithms: the deep convolutional neural network Custom-Net, the MatConvNet algorithm and the Microsoft Oxford Project API.
Face parameters for assessing the similarity of pairs of twins
All participants submitted a saliva sample so that their genome and epigenome could be analyzed, as well as the diversity of bacterial populations in the oral cavity. The couples also filled out a detailed biometric questionnaire and provided information about their lifestyle.
More than 75 percent (25 out of 32) of the pairs turned out to be similar according to the estimates of the two recognition algorithms. This similarity is usually observed in identical twins. For 50 percent of pairs of doubles (16 out of 32), the faces matched according to the estimates of all three algorithms. Then only these couples participated in the study.
In the DNA of the participants, the scientists searched for single nucleotide polymorphisms (SNPs) from a list of 4327108 genetic variants. More than half of the pairs (9 out of 16) had a similar set of SNPs. These nine pairs were called ultra-doubles. Scientists tested them for possible kinship and found that the similarity found was not related to family ties, but arose independently.
The 19277 SNPs common to 9 pairs of ultra-twins were located in 3730 genes, for 1794 of which it is known that they are responsible for the parameters of the face. After that, the researchers compared the SNP data with a catalog of associations between genomic variants and phenotypic traits (GWAS). They determined that 113 SNPs correspond to 84 features of appearance, such as cleft palate/lip, eye color, hip circumference, height, waist-hip size ratio, tendency to baldness, bone mineral density and attached earlobe.
The epigenome of the study participants was studied by checking 850 thousand major methylation sites in the genome. It turned out that only one of the sixteen similar pairs (6.25 percent) had the same DNA methylation pattern. And in general, the twins differed greatly in their epigenomic settings. In addition, scientists have suggested that the external similarity may arise not due to common genes, but due to similar trends in facial aging. The researchers used the "methylation clock" (an epigenetic marker of biological age), analyzing 1,379 methylation sites, and calculated the difference between chronological ("passport") and "epigenetic" age. "Epigenetic" age differences were less among 3 out of 16 (18.7 percent) pairs of ultra-twins than among other pairs (p=0.0052).
The researchers studied the microbiome of the twins by sequencing ribosomal RNA from samples of their saliva. It turned out that the species diversity of bacteria present in the oral cavity is similar only in one pair. Therefore, scientists concluded that the twins in most cases do not have a common microbiome.
The questionnaire data compared with genomic data showed that people with a similar face may have a common physical and probably behavioral phenotype associated with their common genetic variants. Behavioral traits, such as smoking habits, diet, and even the level of education, were closer in similar pairs than in different ones (p=0.00006). Taken together, the results showed that common genetic variations are not only associated with a similar appearance, but also with common habits and behavior.
Thus, the team of Manel Esteller found a correlation between appearance and genetic characteristics. But in order to use this correlation in practice, additional research is needed. Scientists expect that in the future they will be able to predict the structure of a human face based on an individual multiparameter landscape. And this method can already be used to more accurately build sketches and look for criminals, to study the origin Homo sapiens, comparing ancient remains, and finding differences in populations of modern humans. Data on the relationship between genetics and facial features would also be useful in medicine: a face could tell about the features of the human genome and help predict diseases.
Portal "Eternal youth" http://vechnayamolodost.ru