Scientists at the University of Copenhagen have developed a new method that will help in searching human genetic information for beneficial mutations from Neanderthals, Denisovans and other archaic humans. Neanderthals and Denisovans went extinct about 50,000 to 40,000 years ago because of a number of reasons such as climate change, disease and natural catastrophes. One of the reasons, scientists believe, that these humans are known to have interbred with modern humans. During the interbreeding, they passed on over 40 per cent of their genetic information — genome — to present-day humans. As modern humans spread across the planet, beneficial genetic information, upon interaction with the environmental conditions, manifested in many mutations. This process is called adaptive introgression. Some of these changes include changes in skin development and metabolism.
However, many such changes are still undiscovered. The new method, which relies on deep learning of computer models, aims to address this by discovering unknown mutations.
Using the method, scientists not only confirmed the previous discoveries but also found previously unknown patterns in the genome images that could be possible mutations caused as a result of adaptive introgression with archaic humans. Two new possible mutations that scientists were able to find using the new method included changes in genetic information related to blood that could affect blood cell counts. Changes include tumour suppression and various neurological diseases among others.
“We applied it to various human genomic datasets and found several candidates beneficial gene variants that were introduced into the human gene pool," said Fernando Racimo, one of the authors of the study, in a statement by the University of Copenhagen. The study was published in the journal eLife on May 25.
The new method, called “genomatnn" also takes natural selection into account in its models. The method operates on a convolutional neural network (CNN) of deep learning techniques used in visual recognition.