[Science] This is almost certainly not what Denisovans looked like – AI

0
36
[Science] This is almost certainly not what Denisovans looked like – AI


By Michael Le Page Portrait of juvenile female Denisovan, based on DNA analysisMaayan Harel Until recently, the only evidence for the existence of a mysterious group of ancient humans known as the Denisovans was ancient DNA extracted from a fingerbone and three teeth found in the Altai mountains in Siberia. Now a team has created the above portrait of a young Denisovan woman based on that fingerbone DNA – but other researchers are sceptical of the method they used. Like Neanderthals, Denisovans are an extinct type of human that interbred with Homo sapiens. Some people in Asia and Australasia today carry remnants of Denisovan DNA in their genomes, but we still know very little about our ancient cousins. The discovery rewriting the story of humankind: Lee Berger at New Scientist Live Liran Carmel at the Hebrew University of Jerusalem and his colleagues have used Denisovan DNA to generate a portrait that roughly represents what Denisovans looked like. “Our reconstruction is generalistic,” says Carmel. “We just reconstruct the face of the human group, not of a specific individual.” Advertisement Reconstructing faces There’s long been interest in working out what people look like based on their DNA alone, for instance to help identify suspects from a crime scene. But our appearance depends on thousands of variants in gene sequences, each of which usually has only a tiny effect. “Today we cannot predict very much about a person’s bone morphology,” says John Hawks of the University of Wisconsin-Madison. But instead of looking at the DNA sequence of genes to predict appearance, Carmel and his colleagues have looked at how active these genes were. When genes get switched off in cells, epigenetic tags called methyl groups are added to their DNA. The team has developed a way to identify where these tags have been added to ancient DNA. Read more: We may have bred with Denisovans much more recently than we thought The team compared the methylation patterns in the ancient fingerbone to those in bone cells from modern humans and chimpanzees, revealing thousands of genes whose activity was likely different in Denisovans. Next, the team tried to identify which of those changes would affect bone shape, based on what happens when mutations disable these genes in modern humans. Finally, they applied these findings to infer how the growth of Denisovan bones may have differed from ours. The method tells us only in what way the bones differed but not by how much, Carmel says. For instance, it suggests Denisovans had wider lower jaws, like Neanderthals, but how much wider we don’t know. “No, this doesn’t give us any idea of what individuals from the Altai looked like,” says Sheela Athreya of Texas A&M University, of the team’s work. “It’s based on so very many assumptions that it made my head spin.” A partial picture There’s no way to check most of these predictions now but Carmel says future discoveries could reveal whether they have got it right. In fact, after the team submitted their findings for publication, a lower jawbone discovered in 1980 was revealed to be Denisovan. Carmel says its shape matches seven out of eight of their predictions. The team also say they have validated their method by using it to correctly predict some of the known characteristics of Neanderthal bones. The incredible human journey: Alice Roberts at New Scientist Live But Charles Roseman of University of Illinois says their method missed most Neanderthal differences, meaning at best it paints a very partial picture. Hawks agrees. “While this result may seem very persuasive, it is actually not,” he says. “Studying methylation differences is a promising avenue of research, but we are a long way from understanding how differences in methylation may relate to differences in the skeleton.” Journal reference: Cell, DOI: https:// doi.org/10.1016/j.cell.2019.08.035 More on these topics: humans