Joshi et al. reported that look-alike pairs identified by facial recognition algorithms share genotypes but not DNA methylomes and microbiomes.
Based on an earlier study, the authors collected a set of 32
individuals, characterized by their high likeliness, are what are called, in lay-language, look-alike humans, unknown twins, twin strangers, doubles, or Doppelgänger in German [I corrected the German language].
collected by a Canadian artist.
My first expectation is of course that there will be no major effect neither by the methylome nor the microbiome . This seems to be true. So is this a chance event in the genetic lottery or are there cryptic relatedness effects as region of origin is usually shared?
We have about 4,3 million variants on the Omni5 chip while there are 19.277 shared SNPs (in 3.730 genes, the whole data are at https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE142304 and the analysis at https://github.com/mesteller-bioinfolab/lookalike). 19.277/4.300.000=0,45% which can be looked up in an earlier PLOS paper.
14cM/3600cM=0,39% in this table is next to 0,45% so they are cousins of 5th degree – pointing towards cryptic relatedness. The kinship score in Fig 2C does not provide any generation count as the whole paper is more a bioinformatics approach than a genetic epidemiology study (the supplement says 3rd degree which is unrelaible). I would prefer the construction of some pedigrees (or at least using a typical sign like the prognatism) although that might be a lot of work.
We observed that almost all the look-alike pairs cluster close to each other according to their countries of origin (or self-attributed ethnic background) (Figure S3). However, they are not more closely related than other pairs of individuals from the same populations taken at random.
This contradicts not only my estimate above but also Fig 2C. It would be interested to full sequence a predigree and go for shared IBD segments as the crude SNP or gene count ends somewhere in the dust of 158 genes in table S5.
Unfortunately he conclusion of the study is clearly off topic and not justified by any data.
We suggest that these same determinants correlate with both physical and behavioral attributes that constitute human beings.
Behavior cannot be predicted from some facial characteristics. Sorry but we do not need any new eugenics nor any racial genetics as this sentence destroys an otherwise nice paper. Reading it agains, I found one coauthor from a company selling “facial analysis solutions for targeted marketing”…